The Art of Guerrilla Sampling

20150803_152558

Sampling Session for VSCO 2, Summer 2015

Sampling is the process of converting an analog instrument into a digital emulation using recorded “one-shots” of the real instrument in action. Guerrilla sampling is doing all that with minimum costs and maximum efficiency by using existing infrastructure and extreme mobility.

In this post, I will discuss the principles of creating samples, the economics involved, and general tips and tricks to yield the best results when working with time, space, or financial restrictions.

Conceptualization

The first step is to locate and conceive of a library that has value in the market. A bit of basic market research will reveal what sorts of libraries over-saturate the market (e.g. high-powered orchestral libraries, straight-ahead grand piano libraries, acoustic guitar libraries, etc.). Saturation is the approximate ratio of how well the current offerings of that specific product meet consumer demand/expectations. Thus, a fully saturated market has enough if not excess supply of options and features compared to end-user demand.

Another way to tell if a market is over-saturated is to examine the degree of marketing hype on products (i.e. how desperate the company is to sell the products) and to examine the qualities of “industry-leading” product options compared to industry-wide expectations.

For example, if the industry leading Grand Piano library has 32 velocity layers (VL’s), 16 round robins (RR’s), and 8 mixable mic positions, compared to an average professionally sampled instrument which might have 3-5 velocity layers, 2-4 RR (or 8 on staccato), and 4 mixable mic positions, then the piano market is over-saturated.

This becomes harder to measure when the entire industry is over-saturated (as it is rapidly becoming now), as companies try desperately to out-compete in the numbers game (I have seen “indie” developers with 8 mic positions and 16 RR on instruments, for example).

At this time, pretty much everything has some degree of over-saturation- only rare (ophicleide, oboe d’amore, etc.), irrationally expensive (subcontra upright bass, wheelharp, glass armonica), historical (crumhorns, sackbuts, etc.), or otherwise useless instruments remain with light levels of saturation- but also tend to have very low tolerances to saturation (meaning two competing crumhorn libraries would massively cannibalize each other’s sales).

However, that doesn’t mean a type of library which is over-saturated cannot be created and effectively commercialized. The primary method of working in this environment is product differentiation, by which one can present a unique product that covers ground that no other library on the market covers. Product differentiation is the key to creating a unique and effective product in the current sample library market.

Product Differentiation Methods

There are several key directions one can push their library to differentiate it from competitors. I have given these unofficial working names in order to distinguish each strategy-

  • “The Numbers Game” – make the product bigger, bulkier, and more complete than any other solution. This approach is rarely the best, as it tends to result in mistakes in the processing of samples due to the sheer size and difficulty of managing all those samples.
  • “The Technology Game” – make the product more intricate, powerful, and hard to understand than any other solution. This approach is very common in the high-end market, but is very expensive to create due to the immense amount of custom scripting required. Not to mention, this can also lead to confused users who can’t figure out how the library works and end up dissatisfied despite its power.
  • “The Plug ‘n Play Game” – make the product as easy and fool-proof as possible. This approach is shockingly uncommon, and can be highly effective with starting users (a good example is the “single button”-type effects and instruments {embertone sexy sax, for example}, which are virtually foolproof). The downside? These instruments tend to forego on realism and thus have a short shelf-life or limited applications.
  • “The Sketchpad Game” – make the product as lightweight and resource-savvy as possible. In this approach, the product is light on the processor, but via intricate scripts can still be quite decent… sometimes. Other times, it can just be a rough thing designed for composers to toss down ideas on a laptop.
  • “The Emulator Game” – make the product function similarly to a much more expensive product but use cost-saving measures to create an “intermediate” solution that appeals to those who are missed due to that product’s expense (i.e. “the middle ground” approach). This is the approach I advocate personally for all “indie” developers because it is cost-effective and potentially has a large customer base. It can, however, result in a limited product like a plug ‘n play or sketchpad type.

Which strategy is the most powerful? Unfortunately the question is not always the simplest to answer. Different users have different desires and therefore different solutions fit different situations. In current times, the predominant method is the blunt-force Numbers Game, in which, due to the inability of the end-user to distinguish between feature sets as easily as say, different brands of eggs at the super market, the price and statistics become a stand-in for a measure of quality. With this mindset, a product marked at $500 USD but marked 50% off would likely sell much more than an identical product marked at $250 but also 50% off- it has twice the value.

There are few proponents of a strictly Technological approach, such as Sample Modelling, but these products have trouble attracting an audience widely wowed by the number of RR’s and mic positions. However, a large number of traditionally Numbers-based companies are now turning to adding (sometimes useless or pointless) technological features to increase the “wow” factor.

I should also take a moment to note two unfortunate methods of more slimy differentiation that one should be deeply aware of while working-

  • Good Looks – Some plugins rely on just looking good visually. Animated bows, tape decks and fingerings, 3D knobs, something that looks like a video game HUD, 3D rendering of the instrument in immaculate detail. Unfortunately many customers care a lot about how good their instruments look (despite there obviously being no sonic benefit), so take care to at least make your instrument’s UI and marketing material visually attractive and well composed. Consult with or hire a professional designer for best results, but please, for the love of everything good and simple, don’t make it a barfy glitter-fest- unless it’s April Fools or intentionally ironic.
  • Marketing BS – Some companies rely strictly on product descriptions to sell their products (Spitfire is just as guilty as Garritan at this). Customers are generally gullible; when they hear the instrument is sampled at 88.2/24 at Abby Road on a tape deck- despite the painfully obvious flaws with that signal flow to anyone who has preliminary understanding of digital/analog audio and its conversion, will go utterly crazy for it.

Try to be as honest as possible about your product and avoid hyperbole. Don’t brag, describe and let the audience make up their own minds. Intelligent customers will like you all the more for it!

Product Differentiation at Work

Let’s say we have chosen our desired product, but found the market is heavily saturated. It us our task to determine how to best execute the sampling, scripting, and marketing in order to differentiate the product as much as possible from competitors.

Let’s take for example a piano library. Here are a few permutations that would likely be successful-

  • Release a very affordable, lightweight, simple, but very good sounding piano for composers to use in sketching- even possibly offering it for free or as donationware. This is using a combination of the “Sketchpad” and “Plug ‘n Play” approaches.
  • Release a medium-priced but tech-savvy piano featuring alternative tunings and temperaments at the press of a button, some unusual preparations, great built-in effects, and a generally very good overall tone. This is using a combination of the “Technology” and “Emulation” approaches.
  • Release an expensive but extensively sampled rare/valuable instrument with 8 mixable mic positions, 16 dynamic layers, and 8 RR’s, chromatically sampled with both damper on and off. This is a “Numbers” approach.

Currently there already is a piano library that is pretty much complete over-kill for brute force (Vienna Imperial, which took a month or two to sample), so a purely Numbers game is not likely to win out without some fantastic tone behind it. Rather, some degree of technology or straightforward user-friendliness is likely to work best. This measure of effectiveness of each approach differs per instrument and the current market of the specific instrument desired should be studied first before any planning is done.

Planning

There are a few considerations that must be attended to before one can step foot in the recording studio, concert hall, or apartment with mics in hand: a relatively detailed plan of exactly what will be sampled, at what level of detail, and with what end goal must be prepared and committed to memory.

What to Sample

The first consideration is the what of the sampling process. Despite seeming outwardly painfully obvious (It’s a “Piano”/”Oboe”/”Snare Drum”, duuuh!), one must take into account the specific instrument desired-

  • Brand/Make/Model
  • Age, Condition
  • Performer, Performer Ability/Level
  • Location and Portability

For example, one might desire a Mason & Hamlin upright, between 50 and 100 years old, played by a pre-professional (college student) performer, located at a local university or friend’s living room that can be recorded without having to be moved.

That’s pretty darn specific- in most cases you will probably just want an oboe or a snare drum and know you want a competent player- that is, if you aren’t playing it yourself. Often times, the make and model of the instrument is determined strictly by budget and availability. If you get a clarinetist with a professional Buffet Bb clarinet and a second one with a student plastic instrument from China, that’s what you’re sampling (I can’t imagine the former would be too thrilled to play on an unfamiliar student plastic model, or let the latter borrow their expensive instrument and reeds “for consistency!”). Age and condition are often difficult to control unless working with a collector who has several instruments to choose from.

If you aren’t playing the instrument yourself, you’ll need to hunt down a local player or bring in outside talent. If transportation of the item and/or player to you (or you to the item and/or player) is not possible, then consider a remote recording session, where the player is given instructions regarding the sampling process and left to handle the recording process. See more on this in the section labeled “remote recording” below.

Different players of course have different abilities and levels. Ironically, sometimes it is not the most professional players who make the best samples. A little bit of flaw in the tone can go a long way to creating realism, and the best players are often very busy and/or very expensive- hire at risk to your wallet. Keep in mind also that 8 hours with a pre-professional performer will allow you to sample much more material than 1-2 hours with a top player. Some companies outsource their recording to Eastern European countries, where musicians are comparatively much cheaper than the US and UK. Obviously it goes without saying- pay your musicians a fair and competent wage at all times and, if they are part of a union, obey all union requirements. The average payment for a musician in the United States of America according to the census bureau is $35/hour, but professionals and union members will be several times that, in the area of $80-$200/hour depending on region, instrument, and level. See more on this in the section labeled “musician motivation and performance”.

If you are working with a section or group of musicians, then you must always work with at least three musicians, as that is the amount required to form a unified section sound according to acoustical observations regarding the combination of discrete sound sources. Contrary to popular belief, increasing the number of musicians will NOT make a stronger sound, but rather a thinner, less-defined sound. Therefore, it is typically best to stick around 3-7 instrumentalists in a section. If a bigger sound is desired, a second set can be made by either recording a second time and repositioning the players in post, or physically moving the players and layering- time permitting. If no compromise is permitted and a large section is needed (e.g. strings), then use a large section, but consider multiple close mic arrays and/or recording a separate smaller divisi section to reduce the “101 strings effect” that so many orchestral libraries in untrained hands are guilty of.

There may be times or instruments at which it is, by economic urges or skill, more prudent to perform upon the instruments yourself alone. This kind of sampling is slightly tricky to anyone not used to recording alone and can result in either an almost zen-like catharsis, or dreadful rage and frustration as one battles through the 7th hour of isolation with nothing but overtones as companions (results may vary). See more on this in the section labeled “solo sampling”.

How to Find Musicians

If you are at or near a university, even one that doesn’t have a prominent music program, then you will likely have few issues finding musicians. Faculty at high schools and colleges are typically excellent performers for this kind of work, as are professionals in your area. In addition, students themselves are often quite good for sampling projects, particularly if you are near a prominent music college. International schools provide a range of unique instrumentalists on instruments you’ve probably never heard of.

The first step to getting musicians is to locate and attend any local musical functions at which the sorts of musicians you want to sample might be found. For example, a concert of world music at a local university when you want to sample a Guzeng is an excellent start- even if there isn’t a guzeng player there, you might find a dan tranh player or a saz player and want to go in that direction instead.

It’s always a good idea to get a circle of “contacts on the inside” who know lots of musicians. Band leaders, touring musicians, regular session players, and agents are great for getting recommendations. Plus, if you can get a prominent and respected player to work with you, you can always write “Hi, so-and-so said you might be interested in working with me on this sampling project,” essentially lending their legitimacy to your project. These contacts are invaluable and essential to getting large numbers of musicians onboard with your project for things like sectionals.

If you’re a gigging musician yourself, your first contacts might be the people you play with a lot and are on friendly terms with. Your goal should be to make the sampling process fun and appealing to these people- if they like it and think others will like it (or just really like working with you), they will recommend other people to work with you. Pro tip: if someone continually can’t help you due to “conflicts”, there’s a good chance they either just don’t like you enough or don’t have a very high opinion of your sampling project. Most professional musicians are very busy, but if they really like working with you, they are typically happy to throw down a thing here or there in order to help you reach your goal. These are the sorts of people you should try to surround yourself with- as they help you locate players, you can help them back by giving them paying gigs and recommendations in return. The cycle continues, bringing everyone closer to prosperity.

I work off and on with a fantastic player, and over the years he’s helped me put together some fantastic sections for sampling. When he asked if I could play in a group of his where he desperately needed an instrument I can play, I didn’t bother to ask what the price or details were- it was the right thing to do after all the help he has given me, and in the future I could imagine if I need more sections put together in the future, he would probably be more than happy to help. That’s the sort of attitude necessary to build and maintain these connections.

Write to any local faculty via e-mail and ask to make an appointment to discuss the sampling process. If the local high school or college is interested, they might even allow you to use their facilities and property as well as performers, and probably for a lot less than hiring a big studio for weeks on end. Being on good terms with local private lesson teachers, faculty, and professionals is very useful. If there are any instrument collectors in the area as well, make a concerted effort (no pun intended) to meet them and get on good terms. Collections can make fantastic libraries and come with a pre-existing value from the value and unifying factor of the collector and his or her collection.

Through all of this searching, you will likely find a number of musicians, particularly older ones, who are uncertain of or in some cases paranoid about the sampling process. A lot of hard working musicians feel their jobs might be taken away from them by sampling, which definitely sounds threatening enough! Luckily, there’s both a reputation to and an intangible quality to live performance, and in some cases- particularly with uncommon instruments- sampling provides a way to revive the instrument and improve gigging opportunities for the player. You can also always say that your library (hopefully) will force players to write more idiomatically, reducing the amount of bad, boring, or just plain impossible parts players have to read. We all hope that is true, at least.

You should have a 1-2 sentence pitch on exactly what your library is ready to go at all times if someone asks. Consider what your ‘differentiating factor’ is (i.e., what sets your library apart from what already exists?), and consider the connections between your choices of instrument, details, location, and direction.

Where to Sample

There are a number of factors that are behind the location you will be sampling in. The most important factor period is noise. You MUST pick a location which has a low amount of ambient (“white noise”) and impulse (the train going by) noise sources. Of course, when you get to the location, there are a number of important steps you can take to mitigate noise that will be discussed later, but picking a good location can help a lot. Avoid locations directly overlooking city streets. Favor locations facing alleys or in rural areas away from major streets, airports, and railways. Newer buildings are often quieter due to better insulation and construction, and brick buildings are almost always better than wooden buildings at blocking outside sound (if the windows are good).

Studios are of course the instant place of choice in the minds of most, but for budgetary reasons, they may not be the most economical- a studio costs anywhere from $50 to $500 a hour depending on the size, quality, and equipment involved. If you can locate a friend who offers a “friends and family plan” or a local studio that has lowered “off-peak” rates for extra hours or during under-booked periods, don’t hesitate to inquire. If you have your own recording space or a large, treated space, that will work as well. Local schools and universities (nice big auditoriums and noise-treated band rehearsal spaces!) may be willing to allow usage of their spaces and even equipment for free or a small fee during off hours or summer break- a great reason to be friendly with local faculty. Lastly, churches are often in need of donations and will often be willing to allow you to record in their spaces. However, churches can be quite noisy as they don’t typically focus on noise isolation like a school auditorium might, but typically have very good acoustics.

When recording in a new location, take time to scout out the location and inquire BEFORE making any agreements. Spend at least 15 minutes in the location walking around, sitting quietly to hear the ambient noise threshold, and inquire about the noise level of air conditioning, whether it can be disabled if needed, and about any local noise sources of concern. Ideally you want a space that has at least one “quiet” side or corner, around which you can focus your recording efforts, and you want the ability to control or disable any noise sources (A/C, refrigerators, TV’s, etc.) on the day of the sampling session. There is NOTHING worse than showing up at a session with all your gear in tow and realizing it is just too noisy due to loud neighbors to record samples.

Different instruments obviously respond better in different spaces. The goal of the sample set should always be to provide the greatest flexibility to the end user whenever possible (hence why recording multiple mic positions is, at least for commercial libraries, essential). Some users may seek to isolate a dry, close sound, while others may seek a distant wet sound. It is always possible to increase the reverb on samples as a built-in reverb in the sampler, but very difficult to remove unwanted reverb, so always err on the side of less room tone if possible. A particularly active room should not be used for heavy-transient instruments such as drums. Categorically avoid small, ‘active’ rooms, which have many standing waves and will sound boxy, especially with brass (you know the type; those blank drywall-coated spaces you see in new houses these days). A larger space is generally, but not necessarily better as the standing waves will be less annoying.

How to Sample

The next portion of the plan is the numbers- how many round robins, velocity layers, mic positions, intervals (pitches), and articulations. These choices are directly related to the amount of time, and thus capital, available to complete the recordings. One must balance and make an estimate of time involvements in order to determine how much of each of the five detail factors is allocated.

Round Robins OR Multisamples OR Repetitions– repetitions of the exact same velocity layer and articulation parameters to get recorded variations. These variations are particularly crucial with short articulations such as spiccato, pizzicato, staccato, as well as attacks on sustaining notes. Sustaining articulations benefit greatly from 2-4 RR’s, while short articulations should have between 4-16 RR’s available.

Velocity Layers– the number of dynamics isolated by the sample library. It might just be pp and ff, or pp, mp, mf, and ff… or more (to a maximum theoretical of 128 discrete dynamics). Despite common sense, more isn’t always better, as with crossfading instruments, sometimes the best results can be achieved from between 2-3 velocity layers (such as Spitfire’s famous Albion series). A good range is 2-5 for sustaining instruments and small percussion, 4-8 for keyboards and mallets with dynamics, and 6-16 for membranophones (drums) and gongs/cymbals.

Mic Positions– instead of mixing down all the mics to a standard stereo output, each microphone array may be mixed down to a preset position, e.g. close, mids, outrigger, etc. Many products still only offer a single position, but the demand for multiple is quite strong and can serve to greatly help a product’s value.

Intervals OR Pitches (Pitch Fidelity)– instead of sampling every single note on the instrument (chromatically), the sample recordist may choose to sample every other note (wholetone) or every third note (minor thirds). In some cases, other formats, such as the “Diatonic Thirds” method employed on VSCO 2 (in order to match the best resonances on wind instruments, those most closest to the open form), or even in extreme cases, just the tritone cycle, may be used to further reduce time required while retaining respectable detail. While chromatic sampling is lauded as a mark of excellence, wholetone sampling is still widespread, an diatonic is not overly uncommon either (especially since it is easy for musicians to play in tune and comfortably with a good tone on wind and string instruments).

Articulations– discrete techniques to performing on an instrument. It may be as simple as “arco vibrato” or “pizzicato”, or as complex and bizzare as “extreme sul ponticello over Mongolian throat singing”. More articulations is almost always better.

In addition to the above, one can also keep in mind the option of sampling ‘true legato intervals’ for all the intervals on the instrument (chromatically). This method allows the construction of a ‘true legato’ instrument (for sustaining monophonic instruments only)- some even go so far as to record multiple lengths or varieties of legato, and one must also remember: each velocity layer sampled requires a true legato interval set to match for best results.

In the end, one must balance out all factors to create a realistic instrument in the time given. Ideally, one would choose the highest possible of all factors, but on a budget or in tight time, certain reductions must be made in order to fit. To determine which is the first to go, determine if articulations or the interval of sampling is more important. Often times, a wholetone instrument sounds just about as good as a chromatically sampled instrument, while saving 50% of time and 50% of size on the system. Articulations are usually a big selling point, so it’s typically not a great idea to reduce them if possible. In addition, RR’s may be reduced, but for a professional instrument, you want a minimum of 2 RR’s on sustains and 4 RR’s on shorts. Mic positions do not take up more time, and thus it’s best to record as many as you can rationally devise- it’s not hard to ignore or merge useless positions later if they take up too much hard drive space. Lastly, velocity layers can be reduced, especially near the edges of instruments’ ranges, where there is almost always less dynamic range. However, a minimum of three is typically a good idea.

Consider that the process of sampling is effectively a process of ‘analog’ to ‘digital’ conversion. You are taking something which is analog, that is, infinitely variable, and converting it to something that is digital, that is, in discrete, stepped levels. When a musician plays, they are perpetually varying aspects of their performance- volume, timbre, pitch, vibrato, even things like bow position factor in. At the end of the day, it is up to you to take all those disparate variables and assign a set number of “accepted parameters”. Do you record with and without vibrato? Or with three levels of vibrato? Is there tremolo, and how fast is it?

The final decision one has to make regarding how to sample is the order. There are two primary schools of thought:

  1. Emphasize intonation: cycle through all the articulations/velocity layers/RR’s on each note before proceeding to the next; rinse, retune, repeat.
  2. Emphasize consistency: cycle through all notes on each velocity layer of each articulation before proceeding to the next layer.

As may appear evident, these two methods are somewhat exclusive of each other; that is to say, if you emphasize the intonation by recording one note at a time, you may end up with some inconsistent velocity layers if you have too many of them. However, this also means you will save a ton of time when recording a group (particularly the larger the group gets) by not having to retune constantly before every different note take. Therefore, it’s typically best to go with the first “school” when sampling ensembles and may be used for solo instruments.

On the other hand, if you emphasize consistency, if you are working with movable pitch instruments (aerophones, chordophones), pitch may be quite inconsistent to the point that a forte sample might be 10-20 cents sharper than the piano sample (this may be mitigated by using a tuner, but that can cause unnecessary stress and an inferiority complex to surface in the musician). However, this method helps keep instruments much more consistent across velocity layers, and thus is almost a given for all percussion and keyboards, and may also be used for solo instruments.

You may decide to go a completely different route with order, and that’s fine too. Find an order that makes sense to you and is easy to remember and easy to explain (or else can be easily notated out for others if needed).

Microphone Positions

Virtually all standard microphones are mono, meaning they record a single noise source from a specific point in space. Mono used to be acceptable for samples, but is now quite rare- only a few boutique and specialty developers still use mono, and that’s only because they use powerful convolution reverbs to re-create stereo space. Instead, we must use sets of microphones, called Arrays, to create stereo positions.

IMG_1551

Spaced Pair

The most basic array is called a Spaced Pair, or A/B. This setup consists of two microphones placed anywhere from six inches to four feet away from each other roughly perpendicular to the sound source. The resulting difference between the two, called phasing, will result in a stereo image that is quite powerful. Increasing the distance between the two will increase the size of the stereo image. A larger stereo image is more delicate and thin, and as a result less powerful and focused. The obvious drawback of this is that IF someone were to sum the pair to mono, phasing would occur, resulting in possibly undesirable modification of the tone. To pan or adjust the position of the instrument, turn the entire array so that the player is now to one side, thus slightly closer to one mic than the other. This is the same principle used in human hearing to discern position- the difference between the arrival of the sound in each microphone, measurable in milliseconds, can be discerned to an accuracy of up to one degree. Nearly any type of microphone can be used in this array, but I prefer large diaphragm cardioid condensers, which reject sound to the rear, thus improving isolation and reducing unwanted noise, and can have incredibly low self-noise.

The second array is called XY. This array consists of two cardioid (typically small diaphragm condenser) microphones placed in a cross pattern with 90-135 degrees of separation. The diaphragms (the mesh heads at the front) should overlap as close as possible, thus creating a “V” shape, with the arrow pointing towards the sound source. This type of array is called near-coincident, meaning the sound source arrives at roughly the same time to both microphones. Thus, the difference in the signals is primarily what creates the stereo field. An XY array is focused and clear.

XY has a “dark twin” called Blumlein, which can be achieved using figure-8 pattern microphones instead of cardioid/supercardioid, conventionally at 90 degrees separation. Blumlein has no rear rejection, but is a super fascinating pattern because it widens the sound source significantly, while also being collapsible to mono. Essentially, whatever is in the front 90 degrees is spread over the entire listening scope- a source 45 degrees to the left becomes hard left; a source 45 degrees to the right becomes hard right. This is fantastic for making a section sound big and beefy (especially when using ribbon microphones), but can be problematic for sampling as any movement by the player can be multiplied massively if the mic is close. Blumlein actually closely matches the conceptual behavior behind pan knobs, relying predominantly on volume difference rather than timing for the localization of the sound in the stereo image.

The third and fourth types of array is the ORTF and similar NOS. Both again use cardioid or supercardioid (typically small diaphragm condenser) microphones, but this time the rear ends of the microphones form the corner of the “V”, with about 110 degrees of separation at the rear with about 7 inches (17 cm) of separation at the capsules for ORTF, and 90 degrees of separation at the rear with about 12 inches (30 cm) of separation at the capsules for NOS. Both of these approaches are near-coincident, and thus will convert to mono acceptably while having a bit wider of a stereo image than X-Y, but not as spacious as A/B. It roughly approximates human hearing. These arrays are very common in modern orchestral recording, in particular ORTF (though I find NOS is easier to set up on location due to the simple angle and easy distance) and provide a really enjoyable stereo image with consistent, repeatable procedure and results.

There are also Mid-Side (MS) arrays. It is actually not uncommon for these to come built as a single microphone unit with a 5-pin stereo XLR output (which splits into two 3-pin standard mono XLR’s). However, if you don’t have one of these, then you can place a cardioid (or omni, or figure-8) pointing towards the source and a bidirectional (figure-8) capsule pointing to the sides. In order to use the Mid-Side array, you need a Mid-Side matrix converter. The plus side- mid side allows complete control over the width of the sound AND is completely mono compatible, making it quite handy. By increasing or decreasing the volume of the ‘side’ component relative to the ‘mid’ component, you can set the stereo image as anywhere from completely mono/in-phase to completely stereo/out-of-phase.

The final array in common use is the famed Decca Tree. This consists of three microphones- two cardioid LDC’s arranged just like a wide spaced pair, but with a third mic placed forwards by 1/2 the distance between the two LDC’s. Commonly, the spaced pair portion is 2 meters or yards apart, and the forward mic is one meter forward from that center point. The benefit to the Decca Tree is pretty significant- a stereo signal that should theoretically retain its mono compatibility, and it is thus commonly employed in orchestral recordings, placed above the conductor about 10-15′ off the floor (often hanging).

You will have to experiment with each array setup to find which one you like the most. Often times, different arrays will provide different sounds, and the decision may rest solely on the instrument and space you are recording in. In addition, you will have to decide which kinds of microphones to employ.

Microphone Selection

My workhorse microphone was for many years the Rode NT1 cardioid large diaphragm condenser (LDC). It’s very sturdy, affordable (at a very reasonable ~$250- can be replaced or modified if needed without much sweat), and has a natural, fairly flat frequency response that sounds good on just about everything. You can also use more expensive LDC’s, but really the NT1 is a great mic with a solid sound that you won’t feel worried about hauling around the city in a backpack to sampling sessions, not to mention it has ultra low self-noise, meaning it’s not going to add much of any noise to your recordings, unlike many popular modern and vintage mics both cheap and priceless. The NT1-A is brighter and not as well recommended, but still a suitable starting mic. You’re not going to find any mics nearly as quiet aside from maybe the Lewitt 540 Subzero with its hefty $700 price tag and relatively bright sound.

There are a seemingly wide range of other tempting LDC offerings now (note: updating this in 2020). Warm Audio makes some great budget “spin-offs” of popular vintage mics; not quite true to their originals, but still very respectable. Their WA-47jr is by far one of the most affordable yet best sounding “dark” LDC’s you can find. However, their mics are all fairly high self-noise, making them a slightly harder sell for sampling. Similarly, Aston has some very nice LDC’s designed primarily for vocal recording which are rather colorful and very engaging, but again, relatively average noise performance which will be evident if you are in a quiet environment or sampling anything other than loud instruments. It is worth trying out a range of mics, to explore your options, but in general, I would stay away from vintage mics (which are generally noisy and tend to have poor protection from RF and EMF). Transformerless and transformer-equipped mics are both suitable to sampling, but tube mics will generally be on the noisy side and should only be used as close mics or on sufficiently loud sources. If it has more than 20 dB of self noise, it’s totally unsuitable for sampling. Ideally it should have less than 15 dB, optimally less than 10.

You will also need a pair of small diaphragm condensers (SDC). The best budget option I’ve found is the Line Audio CM4 or CM3, with the sE Electronics sE7 tailing behind, but I recommend the sE8 as a strong solution to this with a reasonable price tag, in particular for its neutral tone and excellent noise performance for the price. The Warm Audio WA-84 is not far from the sE8 in character, but somewhat worse in noise performance and at a higher price tag, I find it hard to recommend. Another interesting option is the Blue Hummingbird (in Spring 2020, this is currently available from UK seller Gear4Music for $150/ea), which has a swivel head to help with placement and very low noise, I believe due to a slightly larger diaphragm; however, it is a more ‘colorful’ sound rather than ‘accurate’, and is slightly bright, though I am very fond of them on toms with a little low-end boost. I used to be very enamored with the Rode NT5, but overall it is a bit too bright for most things; on choirs it is exceptional, however, adding clarity and edge to the singing, making it far more engaging; on strings it is pretty atrocious unless you are in a nice big church and balance it out with some more neutral or darker main/far mics. Most if not all other budget SDC’s are wildly bright and papery, or have middling to poor self-noise (smaller diaphragm = higher noise).

If you have no limit to your budget, look for something with a self noise of 15 dB (A) or lower and a linear response; the Josephson C617SET is probably about as ideal of a sampling SDC you can find, provided you have a space where omnidirectional response won’t be problematic… and you can afford the heart-stopping price tag (you’ll need two of them!). Other high end options exist from Schoeps and Earthworks, and are generally greatly lauded although tend to be less focused on noise. Similarly, sE Electronics’ RN17’s are not all that quiet but have been used as close/OH mics on several major pro woodwind libraries. Rode’s TF10’s are designed specifically for recording orchestra (the design process involved famed orchestral recordist Tony Falkner), with a shaped character designed to work great in a concert hall recording situation.

SDC’s are commonly said to respond better to high frequencies, and are thus essential as overheads for recording all forms of percussion, as well as very useful on harps, dulcimers, zithers, etc. However, there are some which are darker and fuller (e.g. Aston Starlights). They won’t sound so great on brass instruments, as they generally lack responsiveness in the lower registers and will thus produce a bit of a “tinny” sound, but can sometimes be used on strings or pianos as overheads to add a little more highs/”shine”.

In addition, I have ribbon mics, a much-beloved Samar VL-373A (stereo, active ribbon) and a pair of Royer R-121-L’s I picked up used. Ribbon mics produce a warm, rounded tone and conventionally have a figure-8 (bidirectional) pattern, which makes them useful particularly on brass and strings, smoothing over the edges and enhancing the bass frequencies pleasantly. A pair of ribbons (or a stereo ribbon mic) can be used to make a spaced pair or Blumlein “mid” in addition or replacement of your LDC mid spaced pair. Ribbon mics are a tricky creature for sampling; while they can have exceptional noise performance with a great preamp and a fairly loud source, with a budget preamp they tend to be very noisy due to having extremely low sensitivity. Some makers have addressed this with active circuitry, essentially a preamp inside the mic that boosts the output.

Ribbons have another significant advantage in that they tend to have very consistent off-axis coloration, meaning when a mic is at an angle to a sound source, it still has a neutral character. This is very much not the case with almost all multi-pattern LDC’s, which almost universally have very colored off-axis response due to their construction. SDC’s, and some (but not all) single pattern LDC’s have much improved off-axis coloration. Off-axis coloration is also problematic in that it is the primary reason why some mics are “easier” to EQ than others. Often times, a very non-linear off-axis response will cause ugly results when an EQ is replied; with a good ribbon mic, you can do crazy boosts and cuts and it comes off as “color” instead of “oh no, someone EQ’d that!” However, sometimes you may actually find off-axis coloration useful, such as rolling off the low end or getting a different character, and in a sense, it is something which is somewhat reflective of human hearing, where the occlusion of our head and ears colors the sound reaching our ears from different directions.

Note that most ribbons, especially budget ones, cater to the “old ribbon” sound- dark, muffled, and honestly somewhat undesirable for most sampling situations where accuracy and transparency is the name of the game. However, for a great budget option, NoHype Audio makes a range of very affordable and reasonably high quality “old-style” ribbons, with fairly aggressive high frequency roll-offs on all of their models. These mics hearken back to the character of vintage mics like

Several companies alternatively produce “modern” ribbon concepts, in particular Samar Audio, sE Electronics, and Mesanovic. The Samar AL-95 is an exceptionally good yet inexpensive ribbon, though it is passive and thus low sensitivity and generally ill-suited to use with entry level interfaces. sE Electronics X1 R is alright, but their VR1 & VR2 are particularly nice and the VR2 is active, though the price is getting up there. At the top end of the “modern ribbon” landscape are the Samar VL-37 and the Mesanovic Model 2. Both are incredibly linear, pleasing mics with exceptional accuracy, though a price tag to boot, especially if you go for their active variants (which is recommended for the lower noise).

Royer makes some very popular, more “traditional”-flavored ribbons. These are color mics almost universally and aren’t nearly as linear or truthful as Samars, Mesanovics, or even sE’s VR mics. It is worth noting that the Royer flavor is desirable in some cases, however, so having some around is handy. Plus, Royers, unlike most ribbons, have a markedly different tonal character between the front and back (I almost universally prefer the back, at least with the 121-Live variant) at close working distances.

Lastly, you will need a variety of close and auxiliary mics:

  • I have a bass drum mic, designed for picking up low frequencies. If you plan on sampling large drums, this is a must. Many typical LDC’s and in particular SDC’s do not have enough pickup in the low and mid-low register that form the basis of a bass drum or large drum’s tone. The Warm Audio WA-47jr is an exception to this, and some ribbons can work very well too, provided they are placed somewhat off-axis and are strong enough to withstand the pressure wave. Place one of these puppies under your taiko, tenor tom, or bass drum and you will be very happy with the result.
  • Lastly, I also carry several extra mics around when the session calls for it- a vintage dictation mic from the 50’s (Shure Hercules), an extra LDC, and a few SM-57’s (or 545SD’s if you’re a hipster like me) for loud things and that classic snare sound (inexpensive and indispensable). Some may even find purpose to have a shotgun mic for recording in particularly noisy environments.

You may even choose to invest in an ambisonic-capable microphone, such as the Tetramic or Rode NT-SF1. Ambisonics has never before (or at least, rarely) been used in sampling, but it is likely to surface as soon as someone builds an ambisonic decoder script for Kontakt- thus ushering a new era of infinite microphone adjustment for those who just can’t get enough of spending hours in their instrument UI’s. However, ambisonic mics tend to have mediocre noise performance, especially those with very small capsules such as the Tetramic, and I have found them to be a poor choice in most sampling situations.

Microphone Specs

When shopping for mics, you should be aware of the following specifications:

  • Frequency Range (’10 Hz to 20 kHz +/- 1 dB’)
    • This specification, commonly quoted, is actually pretty useless, especially without the ‘+/-‘ part. What you should look for is a frequency plot, like this:

      sE8-polar-pattern

      Frequency Plot for sE Electronics sE8

    • Note that such frequency plots are generally ‘touched up’ by the marketing department to look smooth. Also keep an eye on the scale on the left! You know a mic is genuinely smooth in its response when the designer is unafraid to list the smoothing, especially if it’s 1/12 octave (not as much when 1/3 octave…), like the CM4:

      line audio CM4 freq

      Frequency Plot for Line Audio CM4

    • Note that sometimes you DO want a non-linear frequency response! For example, with vocal mics, it is common to have a little boost to the high frequencies, and many mics cut off some super low frequencies to make them audibly cleaner sounding. In other cases, there’s a particular ‘known sound’ common to a certain mic that has been ingrained, e.g. SM-57. It’s definitely not linear, but people like it.

 

  • Polar Pattern (‘Cardioid’)
    • While we all like the shorthand of calling a mic ‘cardioid’ or ‘omni’ or even ‘supercardioid’ and ‘subcardioid’, and most can draw the “idealized” form of each or at least picture it in their head at the mere mention, real mics are not actually exactly the described pattern. Many popular vocal mics are genuinely supercardioid but listed as cardioid, perhaps because the latter is more familiar to a layman. Even if a mic’s polar pattern is true to its shape at 1 kHz, that shape generally changes across the frequency range:

      sE8-polar-pattern

      sE8’s Polar Pattern at various frequencies

    • While the sE8 and other SDC’s tend to look pretty consistent like this, a large diaphragm condenser, in particular a multipattern one, tends to be mostly omnidirectional at low frequencies, becoming more directional and eventually supercardioid at high frequencies, down to a typical narrow lobe of ultra-highs. This is what leads to the phenomenon of off-axis coloration:

      Annotation 2020-07-14 151752

      Aston Spirit’s ‘Cardioid’ Polar Pattern at various frequencies (split lows/highs)

    • Again, this isn’t necessarily bad; you can always turn the mic off-axis to roll off the highs, for example, just something to consider when you are doing things like trying to do XY recording with a large diaphragm condenser and the sound is straight ahead, so at a 45 degree angle to both capsules.
  • Equivalent Noise Level (’20 dB(A)’)
    • Equivalent Noise, or ‘self noise’, is how much noise the electronics and capsule size contribute to the signal before it even gets to your preamp. This is relative to a fixed ‘0’ point for all mics, so you can compare between manufacturers. Most use ‘A-Weighting’ (designed with the ‘A’), which looks about 3-4 dB better than unweighted on paper usually.
    • Atrocious self-noise is in excess of 25 dB(A), poor is 20-24 dB(A), average is 15-19 dB(A), low is 10-14 dB(A), very low is 9 and below. The lowest possible self-noise is about 3.5-4 dB, as you’re actually up against the natural Brownian motion of air molecules at that point.
  • Max Sound Pressure Level/SPL (‘140 dBSPL (0.5% THD @ 1 kHz)’)
    • Max SPL is at what sound pressure level the microphone reaches the distortion level of 0.5% THD at 1 kHz. 120 dB is low, ~130-140 is average, and 150+ is very high. Dynamic microphones have extremely high SPL.
    • Note that 120 dB SPL is still very, very loud; listening to such a sound for more than 15 seconds will damage your hearing. 140 dB SPL is aircraft-engine-next-to-your-face levels of loud. This measurement isn’t terribly important in most recording situations, so long as it’s above 120 dB SPL.
    • In most condenser microphones, SPL is generally determined by its circuitry rather than the diaphragm, though I’ve found small diaphragm mics, many being designed for recording percussion at close quarters, tend to have slightly higher Max SPL than their large diaphragm brethren.
  • Sensitivity at 1kHz into 1 kohm (‘-32 dB re 1 V/Pa’ or ’25 mV/Pa’)
    • Sensitivity seems pretty self-explanatory on the surface. How many volts do you get out for a given sound pressure level? mV/Pa reflects this explanation better, but overall I find dBV to be a more useful measurement, as it’s the sensitivity difference between mics.
      • Sengpiel designates ‘dB re 1 V/Pa’ as Sensitivity and ‘mV/Pa’ as Transfer Factor, which seems fair, but isn’t followed by manufacturers.
      • 1 Pa = 94 dBSPL (sound pressure level)
        • If the transfer factor is 20 mV/Pa, and you record a signal at 94 dBSPL, the mic will show a voltage of 20 mV.
      • dB re 1 V/Pa is basically ‘how many dBV you are lower than if you got 1 volt signal at 94 dBSPL’
        • Generally, if a mic has a sensitivity of -40 dB re 1 V/Pa, it is 10 dB less sensitive than one with a sensitivity of -30 dB re 1 V/Pa.
          • … only so far as 1 kHz is concerned.
          • … generally a mic might be +/- 1 dB from spec.
        • Likewise, if both of those microphones have the same Max SPL of 140 dB, if you hit those mics with a crazy loud signal of 140 dBSPL, you will get +6 dBu and +16 dBu hitting your preamp… the latter being likely to distort the preamp.
      • If you had an absurdly sensitive microphone, like 100 mV/Pa or -20 dB re 1 V/Pa (they do exist!), it would clip many pres by 130 dB.
    • Very low: 2-8 mV/Pa or -52 to -40 dB re 1 V/Pa
      • You MUST have a good preamp OR a very loud signal
    • Low: 10-18 mV/Pa or -38 to -33 dB re 1 V/Pa
      • A decent preamp will definitely be beneficial
    • Medium: 20-30 mV/Pa or -32 to -28 dB re 1 V/Pa
      • Even a basic preamp will do alright
    • High: 32-48 mV/Pa or -28 to -24 dB re 1 V/Pa
      • Preamp is not a major consideration
    • Very high: 50+ mV/Pa or -24+ dB re 1 V/Pa
      • Preamp? What preamp?
    • TL;DR: ,more sensitivity = better for weaker preamps (they have to do less ‘lifting’); less sensitivity = better for louder signals (they generate less signal).
  • Impedance (‘150 Ohm’)
    • The impedance of the mic is generally of little consequence so long as it is under 300 Ohms. Some old mics or ribbons will be at or in excess of 600 Ohms.
    • Ideally, your interface should have an input impedance of at least 10x the output impedance of your mics, so a 600 Ohm mic needs a 6kOhm input on a preamp.
    • Having an incorrect or wildly off impedance will darken or brighten the color of the mic.
      • If the impedance is too low, the mic will also be greatly reduced in volume.
  • Phantom Powering (’48v’)
    • Phantom Power is a 48v DC power supplied by your interface, running down the mic’s cable opposite of the signal. It provides condenser microphones the power necessary to run their circuitry and operate the capsule.
    • Condenser microphones always require phantom power, unless they use a battery or have a discrete power supply which plugs into the wall (e.g. tube mic). Ribbons usually do not, and can be severely damaged by phantom power; dynamics do not require phantom power but generally won’t be damaged by it, at least as bad as ribbons. There are a handful of ‘active’ ribbon and dynamic mics, which can take phantom power, but they are rare and more expensive than their ‘passive’ normal counterparts.

Mic Positions in Practical Use

Let’s listen to an example of mic positions in action. Here’s the microphone setup used, and a few different perspectives/microphones and how they blend:

IMG_1657_smMOD

(sorry, these links are down due to instaud.io closing! will try to fix later)

Up close (on the left), we have:

  • Ribbon mic behind the bell to capture the warmth of tone that the player experiences when performing.
  • LDC and SDC in parallel array (i.e. two mono inputs at equal phase in order to allow the characteristics of two mics to be effectively combined- highs from the SDC, body from the LDC.

Further back (on the right), we have:

  • Spaced stereo pair (LDC’s) to give the stereo positioning and the space of the stage.
  • Far “ambient” mics to give a sound not to dissimilar with what an audience would experience. Far mics are more or less useless to me (as it’s typically more or less “excessive baked reverb”), but some people really enjoy them.

Here’s what the whole thing sounds like as one.

Effectively, this mic setup (7 mics total, 8 if you subbed a decca tree for the spaced pair) allows for me to fine-tune the sound I want from the trumpet… if I want a more orchestral tone, I can switch the far and “main”/spaced pair mics to be louder than the close and vice versa for a more pop/jazz tone. I could export three total mic positions- all the close (as one mono position), the main mics, and the ambient mics, and essentially give this flexibility to the end user.

Of course, this example works great because it was recorded in a great space. This leads to one of the great issues of sampling, which is using your space right. If you are recording in a small studio or room, your primary focus with mic placement should instead be to have as little room tone as possible in the recording. Of course, if we get too close with our main mics, there is too much proximity and the sound tends to become… well, less than pleasant for many instruments (horns, for example, are dreadfully bad close-mic’d), although still quite lovely for many others (pianos, harps, etc.).

A great example of where to use mixable mic positions would have to be, in my opinion, snare drums. Snares are mic heaven, as the chosen picture of the post just comes to show. The colors you can pull and shift through via mixing mic positions are nearly endless.

Here’s an example of a snare drum:

And the whole thing together

Essentially with the snare, we can create an effective close-mic sound by blending the close mics (like so), but also an effective ensemble sound with the spaced pair/ambient mics. This way, the end user has complete control over the setting and placement of the snare and every mic has a distinct purpose in the completion of the tone.

This is where understanding the instrument you are sampling is paramount. Know where the sound comes from, what direction(s) it radiates, where on the instrument it is rounder, in other locations airier, etc. I like to think I know a lot about instruments, but before I sample them I check with every book I have on recording techniques and several orchestration manuals just to make sure I know what I’m talking about and have some rough ideas going in, in case what I have planned mentally doesn’t come out right and I need a new angle.

Also understand your mics. Large diaphragm condensers are the backbone of sampling and a good pair work as main mics on virtually anything. Small diaphragm condensers make great stereo pairs and work well as mid mics, overheads, or even close mics on everything from percussion to plucked strings to even some woodwinds. They have excellent upper detail, but many lack the presence and mids LDC’s so clearly capture. Dynamic mics rarely have uses due to being noisier and typically less clear than condensers, although having an SM-57 for snares or whatnot is always a good idea. Always have some form of bass drum mic (a large LDC or dynamic) for particularly low/loud instruments or to stick under or behind drums or in front of low strings (they excel at picking up that sort of resonance and pizzicato “thumps”). Ribbons- fantastic on brass and gentle on strings, but prone to noise and more fragile than all the other kinds combined.

Audio Interface

An audio interface is a device which generally unifies three historical devices into one:

  • Preamplifier: the preamp takes the mic’s analog signal and amplifies it, raising the volume to the amount necessary to properly gain stage with the converter.
  • Analog to Digital Converter: the A/D converter takes the analog signal and converts it into a digital, usually PCM signal readable and storable by a computer.
  • Audio Interface: the interface allows the signals output by the converter to be sent to your computer, nowadays over USB or Thunderbolt (historically over Firewire or even via a PCI add-in card).

Audio interfaces range from $100 up to $5K+, with varying degrees of quality, numbers and types of inputs, and specs. It is also worth noting that some interfaces, in particular very high end ones, may not have preamps, though generally most do.

Before we progress, let’s take a moment to consider the relationship between the preamp and the A/D converter, as it is very important to recording.

When we put a mic in a room, we’re continuously measuring the sound pressure changes at that point in space. Those pressure changes get turned into a relatively small change in voltage.

The preamp takes that small change in voltage and amplifies it to be orders of magnitude larger. It’s still not much, but at least we can do things with it now! Keep in mind, any noise the mic picked up or generated, plus any noise the preamp generated, is now part of the signal too. If the mic had an initial noise level of “-100 dB” and we add 40 dB of gain, even in a perfect system, the new noise level is “-60 dB”.

However, in most recording cases, unless you are dealing with both a loud source and a quiet space, the noise floor of your space itself, that is, all the things around you (A/C system, cars outside, neighbors walking by, lights humming, the existential screaming of the universe, fridges running, etc.) will make more noise than the noise floor of even your microphone… much less your preamps.

Now the converter takes that signal and at a set time interval called the Sample Rate, assigns a digital value approximating the voltage measured at that point. (it’s a heck of a lot more complicated than that, but this darn thing is long enough as it is…) Consider that the converter behaves a lot like we do sampling stuff: we take something analog and continuous and have to split it up into little discrete bits (well, ah, literally bits in this case). Fortunately for us, the converter is insanely good at its job. However, even the best converters have their limits: there’s a certain point where they just can’t resolve the differences anymore, much like a human trying to read 6-point text.

So, the converter has a minimum level it can distinguish to pick up, while the mic itself is always picking up a level of noise, which is then boosted by the preamp along with the signal. In most typical locations, with a mid to high sensitivity mic, you’re going to be picking up room/environmental noises before you get even close to converter noise.

That doesn’t mean good converters and preamps aren’t worth it; for lower sensitivity mics, quiet environments, and in areas like frequency response and phase linearity they are undisputably better; it just means you don’t have to buy the best interface in the universe to make a decent sample library.

Generally, there are three sections of the interface market: low end, from $100 to $400, mid from $400 to $1500, and high-end, from $1500 up.

Low end interfaces, which range from 1 to 8 preamps generally, have characteristically poor noise performance and their preamps are generally weaker, meaning a low-sensitivity microphone is going to come across as noisy. When I was just getting started, we tried a passive ribbon microphone with my Focusrite Scarlett 18i8 (Gen 1); we had to push the gain to its max, and it was horrendously noisy! We tried using an ART Tube MP, a little $50 tube preamp you can get just about anywhere, and suddenly that noise cleared away. The Tube MP was built to handle the kinds of low sensitivities and high impedances of ribbon and dynamic mics and is able to provide lower noise in such cases.

You will find incremental improvements in each price range of interface in terms of the abilities of the built-in preamps and the A/D conversion. The preamps in my current high end ‘daily driver’, the Antelope Orion Studio 2017, effortlessly provide gain to even my passive ribbons without going too far up the knob, and the converters are experientially about 8-12 dB quieter than the ol’ Scarlett.

High-gain, low noise preamps aren’t necessarily a cure-all either; set up your audio interface and plug in a mic like you’re going to sing/talk/play into it and open your DAW like you’re about to record. Set the right level by playing a little and adjusting the gain until it’s 8-12 dB below clipping on the meters in your DAW. Now be EXTREMELY quiet and keep watching the meter. Where does it flutter around? -40 dB? -50 dB? Maybe even -60 dB or below? Even the cheapest pro interfaces’ A/D converters sit around -90 dB on meters. Best case scenario, top-shelf pro interfaces top out around -112 dB on meters.

However, that doesn’t mean a pro interface is useless; it is entirely possible to record a dynamic range over 90 dB when recording very loud things like percussion, brass, or close mic’d amps. On top of that, better converters can help if you set gain drastically too low, recovering more quiet details. Preamps in pro interfaces also provide significant benefits with high-impedance or low sensitivity mics as discussed previously.

Interface Recommendations

For starting out, something like a Focusrite Scarlett 2i2 or 18i8 is a solid starting choice. There are a ton of nearly-identical interfaces in the budget price range. I would generally advise buying from a reputable, known brand: Behringer, MOTU, PreSonus, Tascam, Audient, SSL, etc. Compare the features, prices, and reviews to pick the device that makes the most sense for you and your current setup.

For something portable, that can be used without needing a computer, you can use a portable recorder such as the Zoom F series, or the Sound Devices MixPre or MixPre II series. All of these also can be used like interfaces, plugged into a computer.

The next step up would be something like a Focusrite Clarett. Above that are various models by Universal Audio, Antelope, who also have higher-end products, joined in that area by companies like RME, Lynx, and others.

Advanced Note on Preamps

While it’s convenient for me to talk about preamps as ‘noisy’ and ‘clean’ or ‘weak’ and ‘strong’, it’s important to note that preamps are actually massively more complex than that. Preamps delay different parts of the signal more than others at higher gain levels, meaning a cheap preamp with one mic turned to full gain and another next to it at low gain will not be completely in phase with each other across the frequency range, even when you compensate for gain in post (try it for yourself!). This graph from Cranborne Audio shows phase shift due to frequency on their relatively excellent Camden pre; the phase shift values will be massively worse with cheaper preamps, or if you have a Highpass filter engaged on some channels but not others:

4-Phase_Shift-All_Gain_Settings

Cranborne Audio’s Camden Mic Pre: Phase Shift vs. Gain

Note that there is a hard limit to how much gain can help; as mentioned before, every mic produces some minimal noise just from existing (oof), as well as pick up room noise. In a noisy space, the room noise will be significantly higher than the mic’s own noise. Preamps also generate a small amount of noise on top of that, focused towards their lowest gain levels. Finally, converters have a more or less fixed noise floor.

Try an experiment: set the gain of your preamp to 0/minimum possible, and pull up a meter capable of displaying down to the level of the signal here. Chances are the meter is displaying something between -94 and -110 dBFS. Turn up the gain a tiny bit and let it settle for a second (it may jump a little on turning up if the gain controls are analog and non-stepped). It should settle down to pretty close to if not the same. Continue to do so, and eventually the noise will appear to rise! Crank it all the way and the noise shoots up!

Oh no, turning up the gain past [~20-50%] increases noise!!!

Well, actually it doesn’t… that noise was there anyway, it was just masked by the worse, fixed noise of the A/D converter. In almost all preamp designs, the noise actually increases slightly slower than the signal does, meaning the signal to noise ratio actually is better at higher gain.

On the other hand, preamps respond non-linearly with their amount of THD+N (total harmonic distortion + noise, or the summation of all the ‘other stuff’ in your signal). The best THD+N figures are generally at medium gain, about the middle of the dial, but that isn’t always true; again, each manufacturer or design style of preamp can have that optimal gain point somewhere else.

So, we can make a few observations:

  • Preamps perform their best usually around mid to high gain.
  • Don’t be afraid to use the last 25% of your preamp’s gain knob; you will get extra THD+N and less linear phase response, but it’s sometimes better than leaving dynamic range on the table.

However, there are some even more general conclusions:

  • More sensitive, lower-self-noise mics placed as far mics, combined with less sensitive, moderately low self-noise mics close up, should in theory hit your preamp with more even gain levels, having similar phase deviations across your mic arrays and thus blending better when aligned in post.
    • Therefore, the best room mics are moderately high sensitivity, low self-noise.
    • The best close mics are mid to low sensitivity, and self noise is not as relevant.
    • You don’t need a low sensitivity mic unless you’re super close mic’ing something or recording a super loud ensemble.
  • There is an optimal preamp for every given mic & signal level combo out there, wherein the optimal recording level is the sweet spot of lowest THD+N in the center of the dial.
    • Preamps built into budget interfaces are optimal with high sensitivity, low self-noise mics because those mics land right in that sweet spot.
    • High-end preamps don’t suffer from high sensitivity mics, though, while being better adapted to handle low sensitivity mics.

Preparation

If you are not recording at the same place your equipment is stored (e.g. personal studio) OR at a professionally equipped studio, you will need to put together a mobile sampling kit. Here is a checklist of everything you will need for a basic multi-mic recording session, all portable by a single individual.

  • Tote bag containing the following-
    • 25′ extension cord
    • Power conditioner or power strip
    • 3-9 20′ XLR cables (one extra, always)
    • Microphone baskets, mounted on quick releases for speed
    • Power units for audio interface and all other hardware units
    • USB or other connector for audio interface
  • Backpack containing the following-
    • Laptop with recording software; laptop charger
    • Audio interface with 2-8 inputs (e.g. Focusrite Scarlett series)
    • Optional: Compact tube preamp or cloudlifter unit (ART Tube MP: $40 of budget butt-saving)
    • 2-3 LDC’s, carefully wrapped in cloth or bag
    • 2-3 SDC’s, ”          “
    • Auxiliary Mics (optional)
    • 1-2 pairs of headphones (monitoring/checking- don’t leave them at home!)
    • 32-128 GB USB 3.0 flash drive (e.g. Sandisk Extreme- very fast!)
  • Microphone Stand Bag containing-
    • 2-8 microphone stands (get the collapse-able lightweight ones if you want use of your arms ever again with more than four stands, not the classic boom arm ones that weigh 10 pounds each)
    • Any mic stand accessories (e.g. XY bar- save a stand, save an arm)

If you are transporting gear alone and have no vehicle and/or have interest in not wasting time, do NOT bring:

  • Hardware mixing board
  • Rackmount units (unless they fit in backpack or can be safely wheeled along)
  • Expensive microphones
  • Fragile or ultra temperature-sensitive equipment

This setup was designed and honed for maximum efficiency and minimum excess weight. It should be portable by a single individual for at least 30 minutes of travel on foot and can be fully set up in less than 5 minutes with practice.

That’s right, five minutes of setup. If you are used to studio recording, you are probably used to anywhere from 30 minutes to a full day of soundchecks, mic tests, etc. With guerrilla sampling, every minute of wasted time could be another extra half octave of range or another dynamic layer of staccatos. I even recommend “drilling” at least twice a week with setting up and breaking down your equipment and getting levels and mic positions so that it becomes second nature. Eventually you want to be completely on auto-pilot through this so that you can explain the process of sampling to the musician while you set up, but more on that later.

For a more advanced setup, you may consider using some rackmount equipment on a mobile rack case or even mounted on a cart (although be sure to get a cart with large air-filled wheels so it doesn’t rattle and damage the equipment when moving outside).

Recording Software

You should find a piece of software that you are comfortable using that allows you to record multi-track audio. Audacity is the most simple tool that can do this, but instead I would recommend either Adobe Audition or Reaper. Most DAW’s will work fine, but they all have quite a bit more overhead and excess than Audition or Reaper, which are pretty minimal on open.

Set up a multi-track session. Create an audio track for each microphone or microphone stereo pair (remember if you have a three-mic array such as a Decca Tree, it should be recorded as a stereo pair and a single mono mic, but the two can be merged and mixed down later on). Your objective is to get as raw and unaltered of a recording as possible- take care to not use any preamp effects or mixdown several positions to one instead of recording them separately. The more flexibility and detail you incorporate in your original recording, the more options you will have later.

Session parameters should be 44.1 kHz @ 24-bit. You won’t use all of the 24 bits per sample due to the limits of modern Digital-Analog Converters (DACs), but it’s best to spend a little extra space so that if you need to do additional processing on the audio such as denoising or tuning, the values are a bit more precise, which help the effects.

That’s right. You don’t need to record at 88.2 kHz or 192 kHz or 96 kHz or any of that. The sample rate needs to only be double the maximum frequency we need it to represent, in this case 20 kHz, the maximum (for young girls below the age of 4) range of human hearing, plus a ~2 kHz buffer to improve processing (but secretly your interface is *already* oversampling at 88.2 kHz+ and converting back to 44.1 kHz, it’s just not telling you that it is). Chances are, even if you are using high-end mics, they can’t pick up frequencies above 20 kHz correctly anyway! So, there’s virtually no point to oversampling except to waste hard-drive space.

This is just the first step, of course. Eventually each track will be exported out as a separate stem and sent to be cut, or cut by you. Reaper is by far the most efficient tool for cutting samples thanks to its unique advanced keybind options and custom macro-esque keybind functions. There are some excellent videos by Simon Dalzell of Ivy Audio on how you can efficiently process and cut samples using Reaper. Some folks have started using Ardour, an open source DAW, for similar reasons.

Sheet Music and Visual Aids

You will want to have prepared a method for proceeding through your objectives. For example, you can work through velocity layers- e.g. recording the entire piano on one velocity layer through all round robins, then proceeding on the next velocity layer. Alternatively, you can work through each note, proceeding through all velocity layers on a single note before proceeding to the next. The first is better for instruments with fixed pitch, such as pianos, where intonation is less likely to change over the short period of time it will take to complete each pass and where the consistency between one note and the next is paramount. The latter is better for instruments with mobile pitch, such as wind and bowed instruments, so the player can really get the note in tune and with good tone (although fatigue may be an issue).

An extreme version of the per-note approach exists where you work through all the articulations on each note before continuing. This is extremely useful for ensembles, where it takes between 15 seconds and 2 minutes to get a single note in tune. You can literally save hours of tuning by getting the note tuned up, then doing all the articulations on that one note, then proceeding to the next.

When using this approach, it is important to make sure everyone remembers where they are in the session- which note, which articulation within that, which VL within that, and which RR within that. It can obviously get quite confusing! Therefore, in some cases, it may be handy to prepare some sheet music that notates each articulation in an orderly fashion with a notation of how many repeats and dynamics is needed.

Working in a Studio Environment

While this text is more designed for individuals working under the radar, there are a number of cases and reasons in which you may be using facilities designed specifically for the recording of audio, the recording studio.

Studios come in all shapes and sizes (literally), but it’s most likely you’ll be recording in a repurposed living space that used to be, or still is, part of someone’s house. As such, the first concern is noise. By default, most studios are selected because they are much quieter and more suitable acoustically for audio recording than other portions of the house or building in which they are located. Custom studios with designer live rooms are specially designed to not only sound great but reduce exterior noise to near silence when closed off- as such, these nicer studios are definitely a great place to record. The problem? If you’re not buddies with the owner, it may cost you an arm and a leg to sample there.

The first thing you should do if you want to record in a studio and definitely don’t want to try the churches, schools, community centers, etc. is to figure out where the local studios are. If they have websites, you should examine pictures of the locations. Look for live rooms (the rooms where recording happens) that are asymmetrical in shape, have high ceilings, a decently large size, and are at least slightly furnished. Even a rug and a bookshelf can make a major improvement in resolving undesired early reflections from the walls, which are the main cause of poor audio quality when recording in smaller spaces.

Also examine pictures for cleanliness of the studios and general professionalism. You definitely don’t want to record for 6-8 hours in a room that reeks of beer and is covered in stains. If the studio appears very busy, then it may not be the right studio for sampling, which requires consistent setups to be retained over several hours or even days undisturbed and without interruption when doing large projects.

Once you have narrowed the search down to 3-5 possibilities, reach out and explain your process and obtain a quote. If you can go by and visit each location, take the opportunity! A classic trick for examining the reverberation of a location is to snap your fingers or clap your hands in various locations throughout the room. What you typically want to avoid is something that sounds like a bunch of people clapping with you or like you’re in a bathroom. Instead, the sound should actually seem to almost disappear and be “sucked up” into the space. It doesn’t need to be completely dry, just dry enough to not have that ‘slap’ effect.

Studios are tricky to prep for because there are so many unknowns. Sometimes they are very well equipped and have so many mics and toys you won’t know what to do with yourself, others, it’s barebones with a few cheap mics. If you can get a glimpse at what their equipment looks like in advance, you can come up with a solution to adapt your procedures to fit their mic locker. For example, you can sub out a LDC you commonly use for another LDC with a similar response and pickup pattern, or bring your own. Don’t be afraid to incorporate the owner of the studio and the engineer in the decision making process. You’re the sample library developer with the end vision in mind, but these guys know this space and gear really well.

You should always try to prepare the engineer and/or owner for exactly what sampling entails. It’s long. It’s slow. It’s kinda boring to watch. They really should just go get a book or take a nap or something, honest. There will be like, two times, when they need to get up and press stop or record. The most complex thing the will be asked to do is glance at the volume bar and make sure we didn’t clip on a particularly loud take. Make sure they know you need individual stems of each array and ensure the sound quality is what you want- there is nothing worse than spending several hours sampling only to learn there was a miscommunication that results in a considerable loss of time and money.

Working in a Concert Hall Environment

Concert halls rock. You always feel like you can’t play a wrong note in one, it all just resonates so beautifully. However, they are often a lot more noisy than you think.

Many concert halls have noisy air conditioning units designed for cooling rooms full of people, as well as fans to cool the lighting equipment and power supply. The best option is to use minimal lighting- don’t use spotlights, which often require the lighting power supply, and instead use worklights which run off the normal power grid. In addition, if possible, turn off the air conditioning and any other noise locations. However, sometimes you can’t turn these things off.

Often school auditoriums have large baffles designed to reflect sound from the back of the stage towards the audience. These can be moved and repurposed to block the sound from one portion of the space very effectively. Although this effectively reduces the size and modifies the nature of the space, it definitely helps. I once reduced noisy AC from one side of a stage that we couldn’t disable by about 30 dB just by using these large extendable baffles, allowing us to record quiet instruments safely, even using far mics.

When recording percussion, it’s particularly important to police the reverberant nature of the hall for the best results. I always use a rug under the percussion so that close mics will produce a dry, clean sound. On a stage, most reflections comes from the ceiling and the floor and it won’t be very pretty compared to the rich reverb in the hall itself.

When running a far mic position, don’t worry about getting “balcony” seats or anything insane like that, just run an array from the middle of the center box (the good seats) with a small snake or two long XLR cables. Even a position right at the edge of the stage is typically enough for “far” mics! If you are planning on recording in an auditorium, it’s a good idea to buy a snake (no, not the living kind) so that you have the option to record distant sources without the living hell that is hooking half a dozen 20′ XLR cables you managed to scavenge in the mic closet on location together only to find one of them doesn’t work (it’s just about on par with replacing broken christmas lights in a string that won’t light up).

Keep in mind that if you use cardioid main mics, they will reject a lot of the hall (which is often good unless you like your main mix to sound wet, although that is typically unadvisable). Omni pattern mics will pick up everything in all directions, so they are good for a larger main mic sound.

One of the main concerns of sampling in a concert hall is the difference in absorption between a full hall and, what you will likely (hopefully) be sampling in, an empty hall. Some go so far as to bring in sacks with similar acoustic properties to the human body and setting up a full orchestral setup. This really isn’t necessary, but can be done as a product differentiation method if desired.

The Session

This section is dedicated to working with live performers. If you are planning on recording yourself alone, skip to “Solo Sampling”.

Coordination and Communication

Locate and lock in your performers at least 72 hours (three days) before the session, if not a whole week. Remember professionals book up typically a week in advance if not more, so you should take care to book in advance. Always get the musician onboard first, then get the location. It is way easier to find an alternate location than it is to find an alternate musician in virtually all cases. Clarify that the musician will bring their own instrument, and provide them with detailed instructions on how to reach the location if they are coming to you. If you don’t have a place to record, you may intimate discreetly that you are in need of a common location or “would be willing to come to you if that is more convenient.”

In your communications (and this goes in general), try not to be super serious and business-like, au contraire! You should, while considering the suggestions above, try to be personable and as honest as possible. Address people by their first name and memorize it. A huge part of successfully sampling musicians is getting them to RELAX, which, as a sample library developer, is honestly the hardest thing you will have to do during the entire session.

Musician Motivation and Performance

Everyone gets nervous, even the most veteran performers. Sampling to about 98% of musicians is a new and unusual experience, and no matter how many times you explain it is “no more complicated than playing long-tone warmups” or “the easiest session you will ever play on,” people are still nervous.

Your goal, as conductor of the session, is to knock the musician totally off their guard. It helps if you are already friends, but if you aren’t this is ten times as important.

Why shouldn’t musicians be stressed? Doesn’t stress make people think critically and clearly, and worry about precision and accuracy?

Stress makes people hyper-critical, not to mention causes muscles to contract and the diaphragm to shrink- for wind players, this is not very good, and even for string players this leads to poor posture. Sampling requires sitting for long periods of time in the same place and doing small, precise actions over and over again. Thus, a perfect storm leading to an unhappy musician with potential back problems, poor, weak samples, and a waste of potential.

This is especially true with ensembles. Stressed out musicians will spend much longer trying to tune, get frustrated and dejected much faster from the tuning process, and overall give a much poorer performance.

Even “good” stress, making an individual seem serious and professional, can still cause internal changes that affect diaphragm and mental changes that affect the endurance and most importantly, confidence.

Confident, relaxed musicians therefore make the best samples. Humor is quite possibly the easiest and fastest way to win the confidence of the musicians you are working with in the very brief time you have. Humor and confidence will empower the performers to ask clarifying questions and give valuable suggestions, such as suggesting a change in tuning method that is much faster, or suggesting an articulation that would otherwise go unsampled (*cough cough* sul ponticello over Mongolian throat singing *cough cough*).

Each musician you work with will have a different reaction to the process. Many really open their ears and catch a “glimpse” of the magic deep listening reveals inside otherwise plain notes. Others listen deeply but become very self-critical in the process- every note they ask for a re-take; “it has to be just right,” they say. On the other hand, some people get really, really silly by the end. I once was almost in tears laughing as we tried to sample a few unusual trills on a flute requiring both the player and myself to get to work (the only known instance of flauto four-hands in recorded history).

So where do you start with getting the musician on the right path?

First Steps (literally)

Begin by arriving approximately 5-15 minutes early if you need to pick up a key or get special access to the facility. If it’s a studio, get there 10-25 minutes early to be safe. Otherwise, arrive just a few minutes early. Your goal is to be setting up while the musician is setting up. This might sound inefficient, to wait, but the communal setting up and warming up time will allow you to start getting the musician comfortable. There’s just about nothing more intimidating to a musician than walking into a room full of microphones set up like the skeletons of a half a dozen charred trees and being rushed to set up and warm up.

When they arrive, greet them, shake hands, smile – the usual. If you are typically introverted as is yours truly, try to be particularly outgoing for this bit.

If you didn’t have the opportunity to scout out the space beforehand, immediately begin by walking around the space listening carefully. Locate all noise sources (AC vents, traffic, neighbors, refrigerators)- your first order of business is to determine where to put the musician and where to put the mics. Cardioid mics should always face away from the sources of unwanted noise, and the musician should be placed anywhere that isn’t right up against drywall. If possible, choose an area roughly near the middle of the space.

If there are bad noise sources, seek methods within your immediate area to resolve them. For example, ask to unplug a fridge temporarily, or turn off a loud AC unit. I even once found there was a rodent repellent unit that would emit a loud beep- it would have ruined the delicate harp samples we were recording if it had not been temporarily disabled. I am also infamous for taking battery powered clocks off the wall and removing the batteries just to avoid the risk of the ticking getting in the recordings.

You should already know where to put your mics the minute you figure out where the musician(s) goes. Main mics go about 5-12 feet back from the musician’s seating position, with a close mic about a foot away- consult books on microphone placement or experiment around to place the close mic.

First step is to unpack the laptop and interface and get that plugged in and warming up. Your setting up should be deliberate, but never rushed. Again, rushing or appearing frantic/stressed – unless for clearly comedic effect – will cause discomfort in the musician(s).

When determining location for your equipment, try to keep it off-axis to the recording equipment. Not only can electromagnetic fields from the transformers potentially cause issues with the audio signal (remember, it’s just an electrical current… although XLR ingeniously preserves the signal from EM radiation), but a noisy cooling fan or, even worse, a noisy engineer, can cause unwanted sounds. For that reason, I usually seat myself on a stool (or stand!) right behind the two main mics so I’m off axis but still easily visible by the musicians to give conduction and whatnot as needed.

Objective: Social Mode

Since you walked into the room, you should simultaneously start discussing the project with the musician(s). Always start with the musician(s)- if you haven’t already, ask what styles of music they play (this may reveal additional articulations you were not anticipating or remove some you were), and inquire how long they have played. If they are a student, you can always inquire what path of study they are in and what semester they are. Engaging in this sort of talk engages what is called “Social mode” in the individual.

According to economic theory, humans function under two dominant modes, “Social mode” and “Economic mode”. When a person is talking to or assisting a friend, they are typically automatically in social mode- that there are benefits to engaging in the communication that transcend simple economic terms, that some “investments” of time now might mean a favor in return in the future, etc. However, when with a stranger, the individual is in economic mode- under constant suspicion and vigilance that the other may be trying to give them a bad deal, subconsciously highly aware that the activity engaged in has specific profits and losses associated. Your goal is to get the individual into social mode by connecting to the individual on a personal basis.

Humor is another great tool here. Don’t be afraid to tell a joke, even a really bad stupid one… or even use a silly voice or face for something or shout when telling people to play “forte” and whisper when instructing to play “piano” (channel your internal crazy German conductor). I often ask the musicians I work with if they have heard any good jokes recently while I set up. Of course, don’t let the humor side become too prevalent or else you won’t work efficiently enough (and won’t have enough time to explain the procedure), but it’s a good way to get people in the right mindset: this is going to be relaxed, this is going to be open, this is going to be safe. Introverted and shy musicians require encouragement in particular- an extra minute spent at the start could save minutes later on of bad takes and discouraged resignation.

Another method to help people relax is to avoid “weak” postures, such as crossed arms, slouching, etc. If you use a “power stance” or even just a relaxed but alert posture, the players will take the cue subconsciously. If you have impeccable posture when sitting, your musicians will automatically copy you. If you don’t believe me, cross your legs in a waiting room or on a train car when no one else is doing it and wait to see all the people across from you cross their legs too.

Explaining Procedure

Now comes the tricky bit. Explaining your entire job to this slightly dazed stranger in about 3 minutes without turning it into a confusing monologue of technical jargon. While you do this, you want to be hooking up mics.

Start by telling them what the samples are for especially about how it will help lots of composers realize their works more realistically (good time to throw in a “and hopefully improve those less-than-ideal parts they ask you to play in sessions”). Now you need to explain in a general sense how the sampling procedure works itself- what note they should start on, what scale they should use, how many times to play each note. You can even sing or play on piano (if one is accessible) an abbreviated demonstration of the cycle- this often helps a lot!

If you have a waiver, NDA, or contract you need signed, you should probably do that now. Keep it to one page, and make it human readable and to the point. You may even want to throw in a joke “name your firstborn son ‘Michael Praetorius'” term or something for fun. Remember, a jolt of “super serious” will instantly throw them back into “economic mode” which will undo all of your work so far into getting them to relax.

This is a good time to talk about The Sampling Ground-Rules:

  • Remove any danglies, bracelets, or objects that might rub against the chair, instrument, etc. Don’t wear jackets or other noisy clothing.
  • Take care not to move during the performance of the note. Especially don’t touch or rub/scratch your denim jeans… or shirt… or hair… honestly, leave hair alone.
  • Always, always breathe through your open mouth while playing (obviously not applicable for winds). The nose is very noisy.
  • Hold your position once the note has ended until it has died away.
  • If at any time we need to re-take (likely because something loud and annoying went by on the street outside- GOD how dare those ambulances try to save people!), I’ll let you know.
  • If at any time you would like to take a break or ask a question, don’t hesitate to let me know!
  • Please turn off your cell phones and return the table tray on the seat in front of you to the upright position. Violation of this will result in no fruit-cup.
  • The anesthesia typically wears off within 24 hours.

Always start with a ‘test run’ once you have set up the mics and turned on phantom power where needed. First ask the musician if they would like to play a little something while you calibrate the mics. Ask the player to play the loudest thing they can (sometimes this can be fun to say with comedic effect as well… “You call that loud?”) and set the inputs so they peak between -3 and -9 dB for safety. Then you should instruct the musician through each step of the sequence as a dry run, going at least to the second note in the series (e.g. C @ piano, C @ mezzo-forte, C @ forte; D @ piano, D @ mezzo-forte). Be sure to let them know what is too long and what is too short with both longs and shorts.

MAKE SURE your microphones are set up right- Left should be Audience Left, not Player Left, for all positions. It can be reversed later on, but it’s easier to just record it right to start. I often walk around and speak the desired name of the mic into each mic and make sure it’s set up right while recording the test.

This will be your last opportunity to get inputs balanced- do it while they are playing using the volume bars on your monitor while listening with earphones. Make any final adjustments- remember, your job is to capture a clean, pure sound that can be modified and adjusted if needed. Ensure any tube preamp is not adding unwanted distortion if it is in use. MOST IMPORTANT: Once these values are set, do NOT change them.

Once you have a good balance and sound and the musician seems comfortable with their first sampling task, press record and walk them through the first few bits.The first few notes are often reluctant and feel unnatural. You can see this on the faces and in the nervous laughs of many musicians who play for you in sessions. But, as time goes on, after about 10-15 minutes, a pattern evolves. A systematic, organized process takes over. Eventually they’ll get the hang of it. If they still seem unsettled or uncomfortable, if a take really sounds great, let them know! A little ego fuel goes a long way.

Special Considerations/Tips

Certain instruments require special considerations. Here are a few more common ones to keep in mind-

  • As string players work up the fingerboard, depending on the height, they will become fatigued if they have to hold down a string for a while. Give them the opportunity to stretch and take a break regularly.
  • All strings will have unwanted resonance of the open strings. Unless you are sampling chromatically, you must mute the other open strings or else it will sound really, really wrong.
  • With most wind instruments, it’s a good idea to work up from a low or middle note, such as the low C on trumpet (transposed), then to come down and ‘pick up’ the last few low notes at the end. This way the ends of the range can be worked up and down into, rather than instantly lept into, allowing the player to ease up or down.
  • Trumpet players (and by extension, all brass to lesser degrees) fatigue quickly, particularly on high notes. Be careful to offer them regular breaks so they can keep their chops healthy.
  • Sometimes you need to just say “let’s take a break” when a player is starting to get discouraged with a lack of control (sampling often makes players hyper-critical and also unearths a lot of previously unknown control issues which virtually all players experience).
  • Sections require extra time to tune. Therefore, it’s best to sample per-note- get the note in tune, then play all the articulations and varieties of that note, then move on.
  • Instruments such as harps, pianos, and harpsichords all require at least 5-15 minutes of tuning before sampling can proceed- see if that can be done for you by a professional tuner before you even enter. Harps are fairly quick so the harpist can typically tune up while you set up without any delays. Never sample an instrument with more than one string that is played in unison (piano, hammered dulcimer, mandolin) without tuning first.
  • When placing mics, contrary to popular opinion on mics (less is typically better), it can’t hurt to place a few extra close mics to get extra perspectives. It really sucks when your piano close mic doesn’t sound the way you want it to sound because you only used one close mic. Additionally, recording an audience perspective, despite being pretty worthless for anyone with even a remotely decent convolution reverb plugin, is a good selling point for the “numbers” game.
  • Spend at least 30 minutes of the session doing something wacky or non-standard. You will thank yourself later when a reviewer says, “what a cool sound!” every 10 minutes.
  • Horns are VERY hard to tune. Don’t sample more than four of them at once if you want to keep your head and schedule. They also take a long time to empty out condensation, so the more horns, the more delays.
  • Horns sound terrible in anything other than a concert hall or other large acoustically pleasing space. Do not sample horns in a small studio or classroom, it will sound like a 1980’s sampler horn no matter how detailed you make it or what mics you use. Oh, and don’t put a mic right behind the bell. Just use an ‘overhead’ slight above and in front of the player- sounds much better.
  • If you want your horns to project more, put a sideways table or other flat surface behind them to reflect the sound.
  • You can automatically instill a sense of placement in your samples if you slightly turn the main and all subsequent arrays slightly the opposite direction you want to move the players  (i.e. to move players clockwise, move array counterclockwise) or else move the players chairs physically around the space.
  • All keyed instruments (e.g. clarinets, saxophones, flutes, etc.) project sound out of the key areas as well as the bell/end. Typically a close mic about 1-2 feet from the instrument pointing towards the center works well enough for starters.

Conduction

If you are leading a section, practice your conducting. You should never need to conduct for a solo player- it will add unnecessary tension and stress. For the love of God, don’t use wavey-smoothy conducting patterns– just use very precise, geometric, and steady patterns that show each beat with a clear and pronounced ictus (the little bounce on the hit). Hit every beat in distinct locations so that the players, upon a glance, can see where you are, and exaggerate the “4 +” before every measure. Also practice preparations (the “4 +” you give before starting a piece) and cutoffs, practicing precision and clarity on these as well. This abbreviated, militant conducting style (“a la band director”) is necessary to provide the clearest, most accurate timing possible when leading a section.

When playing shorts (staccato, pizzicato, etc.) it must be conducted. Have a 4/4 or 3/4 pattern with a note on the first beat of each measure. Practice and rehearse a standard signal to proceed to the next velocity layer and instruct players to continue repeating the parameters until instructed otherwise- there will always be mistakes during recording shorts, and it gets worse the more players you have. During beat 3 or 4, once the sound has died away, you may deliver brief verbal cues, such as “forte”, “staccato piano”, and so on.

There is no reason to use a click, but keeping a steady tempo is necessary. Determine a tempo by finding the decay of the sound of the instruments and adding some extra buffer, then using either a 3/4 or 4/4 pattern depending on how slow you have to go. Don’t use fancy patterns or overly complicate things, just focus on making the basics as clear as possible.

Other Methods for Inspiration

Sometimes musicians need a bit more of a nudge to relax. One of my favorite exercises is something I picked up when I did free improvisation regularly- a listening game. As they hold a note, ask them to listen to the overtones and the way they interact. I was once in a session with a bassoonist and we realized there was a high overtone that was really strong. Of course she had been playing bassoon for decades and never realized this about the bassoon!

People sometimes like to hear themselves- sometimes it can help to play them back the first pre-take.

Another thing that can really help for groups is to ask them if they know a song or to invite them to improvise. Sometimes I’ll do a quick 1-minute free improv. at the start. Record it and splice up bits to create some aleatoric samples! Other times, I’ll bring some super simple harmonic music so players can get used to playing together. Most soloists have their own repertoire that they like to play while warming up.

I’ll also use an aleatoric free improv activity as a warm-down as well, right before the end of the session. Simple conduction exercises- little 15-20 second “vignettes” of harmony. Have several be with the feeling “quick” with short notes and another “long” then conduct an attack, growth through the phrase, then a slow decline back to silence. These go excellently in the library and are always a lot of fun for players, especially more classical conservatory-trained players who rarely get invited to experiment on their instruments like that.

The End/Compensation

You will often need to set aside ~10 minutes at the end for packing up and paying musicians. With mobile deposit methods, checks are an effective way to pay musicians if you have qualms with carrying around large amounts of cash. In addition, some musicians may prefer to be paid via paypal. Wire transfers are rare and typically not cost effective unless the sum is large. I typically offer the musicians a choice between a check on location, cash sent in the post (or sometimes on location), or a paypal transfer that evening, and collect the relevant information as part of the release they sign at the start.

Please note if you are paying someone over $600 here in the United States, you will need to collect a W-9 from them and provide them with a Form 1099 at the start of the next year. Talk to your CPA or financial adviser for additional information on this.

Snacks and Beverages/Alt. Compensation

It’s always a good idea to have water available for musicians during the sessions. If the location allows you to bring food, then you may want to consider bringing some food for the musicians for the end as a little celebratory gesture. Musicians love food, especially if it’s something delicious and home-made. Look up “graham cracker pralines” for a simple and absolutely deadly dessert I often bring to sessions (make it with dark brown sugar for extra fun, and make sure you break up the graham crackers into individual tiles first with a knife).

There may also be cases where a musician might actually be interested in food as part of their compensation, and this can be a cost saving opportunity for you. For example, instead of paying $40/hour, one could pay $20/hour and make a gourmet home-cooked meal for the players afterwards. Even fairly involved multi-course meals often don’t cost more than $10 per person if you prepare it all yourself, so you would theoretically save upwards of $10-20 per person per hour depending on the number of hours.

This also gives a great opportunity to cement your relations with the players so that if you want to do future sampling with them, it will be an easy and enjoyable process. Once musicians are onboard with what you’re doing, sampling can be a lot of fun.

Solo Sampling

Sampling instruments by yourself is a fantastic way to get stuff done without all the drama and work required to get musicians motivated. However, it can cause some… undesired side effects if you aren’t careful. For one, long periods of sitting and intense focus can cause stress resulting in disorientation and even acid reflux, especially if you’re playing physically demanding instruments such as low brass or large percussion. One solo sampling session was so long and sustained (about 10 hours with a short lunch “break” eaten on location) that I nearly threw up as a result!

The first rule of solo sampling is to take care of yourself. If you start to get too fuzzy-headed, it’s time to stop and come back later… hopefully. It’s important to thus set up systems to help mitigate stress and make the process comfortable. Studies by the Israeli army showed that soldiers on the move should stop and rest about 5 minutes per hour, with a 15 minute “packs-off” break every 3-4 hours. We should thus assume the opposite- if you sample sitting for a hour, take five (and I MEAN five whole minutes of taking a break- stretching, walking, etc., not the song). After 3-4 hours, take an extended break. Go outside. Take a walk.

The second rule is to plan ahead. Make sure you are maximizing your time if it’s limited. Come up with a series of priorities on what you really desperately want to sample and what can wait or be supplemented elsewhere or by someone else.

This plan relies on four factors, regardless of if you are the one playing or if someone else is playing for you:

  • Scale: How will you approach the instrument? Diatonically? Chromatically? How much time do you have? See this blog post for some ideas and reflections on the matter.
  • Articulations: What are the common articulations? Can you play them? How many velocity layers can you divide it in? Do you need round-robin/multisampling?
  • Range: Can you play the outer range of the instrument? Do you need to adjust your plan to help your endurance?
  • Mics: How are you going to mic your instrument? What tone do you want to bring out? How does the instrument sound in tests?

These four factors all lead into the overall feel and functionality of the final product. Some people like planning these meticulously months ahead of time, others such as myself do some general planning and let the rest fall into place in case there are issues or other concerns (which there always seem to be).

Regardless, you must always plan your time and have a preliminary plan for mics when you go into a session. Range and exact Scale can be determined on the spot, as can some final mic placement changes to adapt to room environment and instrument behavior/design.

Preparing Yourself

Before going to your session, make sure you eat a healthy, full meal. Do not sample on an empty stomach- it is a very bad idea, trust me. I personally find I play better if I have eaten at least a hour before I play but no sooner. Therefore, I suggest eating about 1-2 hours before the session so by the time of the session, the energy is starting to come clean and steady without any discomfort from having just eaten.

Bring at least two full water bottles’ worth of water, even if the location has water fountains or vending machines present. That way you don’t have to get up and stop recording to take a drink. If you plan on being there for more than a two hours, you may want to bring some light snacks (a piece of fruit, granola bar, etc.). There’s almost nothing worse than being starving AND trying to focus intently and perform your best.

You should prepare your mobile device or tablet with two very important apps (or, if you can find one, an app that combines these two elements)- an accurate tuner (I use cleartune but there are a variety out there) and a decibel meter that shows decibel levels like a standard volume bar. If you don’t have a mobile device or tablet, you can use applications or plugins on your computer that are monitoring the input signal live. You will need the first when playing a wind or fretless stringed instrument to get good intonation. It’s entirely possible to change tuning later on, but what you want to avoid is large amounts of pitch waver and having your entire instrument wildly out of tune. In the latter case, you will end up sampling something that isn’t an accurate representation of the instrument, while in the former, you will greatly tick off whoever has to process your samples.

Be efficient and experienced with your equipment. Regardless of your equipment, be it a pair of SM-57’s or bespoke ribbon microphones, it’s important to understand completely the advantages and disadvantages of your equipment. Poor samples almost always result from generic usage of equipment, such as is often imposed or proposed by books, engineers, or teachers. The problem is, most people (and books) got their advice from people who lived in an era when stereo recording was young. We really don’t know much of anything about stereo recording outside of the theoretical and the handful of observed and tested models- XY, ORTF, NOS, AB/Spaced, “Decca”, and MS.

Calibration

The way sampling alone typically ends up is, you spend the first 10-20 minutes setting up and calibrating the microphones to your taste (meaning every time you want to make an adjustment, you have to put your instrument down, go run over to the interface, turn some knobs, and then run back).

When you are initially calibrating, you’ll want to record a short song- I came up with a little one that I play on all the instruments I sample (it’s just a little 4-bar Baroque-ish ditty that repeats itself). Now you can go adjust levels based on the multi-track recording.

This is a good time to talk about The Sampling Ground-Rules for Flying Solo:

  • Remove any danglies, bracelets, or objects that might rub against the chair, instrument, etc. Don’t wear jackets or otherwise noisy clothing.
  • Take care not to move during the performance of the note. Especially don’t touch or rub/scratch your denim jeans… or shirt… or hair… honestly, leave hair alone.
  • Always, always breathe through your open mouth while playing (obviously not applicable for winds). The nose is very noisy.
  • Hold your position once the note has ended until it has died away.
  • Take regular breaks to stand up and walk around. Your legs, bum, and back will thank you later.
  • Please turn off your cell phones and return the table tray on the seat in front of you to the upright position. Violation of this will result in no fruit-cup.

The Marathon

After all that planning comes the first note and “the marathon” begins. You enter a game of mental and physical endurance- the normal challenges of performing augmented by the demands of a careful performance (no giant flubs, please), not to mention monitoring the progress and success of your session (RR? Note? Velocity Layer?)- often with fingers involved. The first few notes are often reluctant and feel unnatural. You can see this on the faces and in the nervous laughs of many musicians who play for you in sessions. But, as time goes on, after about 10-15 minutes, a pattern evolves. A systematic, organized process takes over. Velocity layers start becoming more like emotional states cast over scientific exactness: a forte might be anger at eight inches, a piano, resignation at one.

As human beings, we love to be able to understand our world via classification and organization. We develop patterns to create a sense of continuity and comprehension of our surroundings. While sampling, this natural order asserts itself and the repetition, like a mantra, becomes second nature. You begin to forget that you are playing an instrument. You begin to forget that you are using your fingers, lungs, eyes, and ears to carefully perform and shape each note and slip into a trancelike state, feeling and straining to hear each note dissolve into the beautiful nothingness of the dark hall that lies before you. Sometimes you jolt to your senses realizing you just held a piano key down for a minute and a half, listening to the almost imaginary whisperings of the very last resonances in the strings. Other times you sit there and realize you have been sitting silently for a totally unknown amount of time, so intently listening that even the silence made music.

In many ways, sampling is similar to piano tuning tuning. It encourages a deep listening that most musicians don’t encounter in everyday playing. You become strikingly aware of the shape and the breadth of each note that you produce, and the subtle beating of strings and harmonics and all else. Weird overtones that you never noticed before play sharply in your ears. It’s a fascinating state, and has opened my ears in new ways while performing and composing (not to mention fueled an interest in minimalist free improvisation).

Eventually you find yourself at the end with nothing more to play. You have finished your work, like a marathon runner a race, and now it is time to rinse, wash, and repeat with the next instrument or articulation.

Aftermath

Once you have completed recording on your sampling project, it is time to gather all the finished audio and head home. If you worked with other musicians, you should make sure you collected their contact information so that you can let them know when the project is out- it’s also a nice idea to offer them a copy of the library (or at least a respectable discount for their contribution). Be sure to personally thank anyone who worked on the project or helped you locate musicians, from engineers and facility managers all the way down to assistants and runners. Never burn bridges, even if you don’t think you’ll ever use the facility or performer again for any number of reasons.

The next step for the sample library is cutting, followed by mapping and scripting. These will not be covered in this post for reasons of length and time, but a decent amount of video tutorials regarding the cutting and scripting processes can be found by Simon Dalzell of Ivy Audio on Youtube. If you are not able to cut your own samples, you can find a number of companies, such as Elan Hickler over at Soundemote, who will provide cutting and processing services for a reasonable fee. Mapping and scripting are equally challenging, but there are a variety of choices that have to be made before deciding on the right sampler format for your product, which will inform how you map and script it.

For now, be sure you back up your raw sessions in their entirety to at least one additional backup drive (I have them located on three drives across multiple systems). If you back them up online, also keep a physical backup somewhere just to be safe. If you just spent, say, $500 recording a bunch of samples, then there’s no reason to spend some time and money to keep that investment safe.

If you have any further questions about the sampling process, affordable equipment, or anything contained in this blog, don’t hesitate to contact me.

Best wishes,

Sam Gossner, Founder
Versilian Studios LLC.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s