Once a rock star’s luxury, now an essential instrument in every professional and home studio. We explore and explain the core features of the sampler…
Cast your mind back to 1979 and the early ’80s and you may recall that the instrument of desire at that time was the sampler. The Fairlight and the Digital Synclavier, costing more than most of us expected to make in a lifetime, made us drool as they demonstrated the power of sampling technology.
Throughout the 80s samplers became more affordable (although many still cost a year’s wages) and then, as the 90s progressed and finally turned into the new century, software samplers began to appear offering more power and versatility than their hardware forefathers and at much lower prices, too.
But whether a sampler is hardware- or software-based, they all have a common set of functions and features for arranging and manipulating the raw sample material.
Rates for the job
Analogue-to-digital converter, the gizmo that samples audio and converts it into digital data.
Short for Binary digIT, a number which can only have one of two values – 0 or 1 – and which is used by computers at their lowest level of operation.
Changing the pitch of a sample by playing it back faster (to raise pitch) or slower (to lower it).
Posh word for tone.
A general term for the assignment of samples to keys. It covers key splitting, layering and multisampling.
Analogue-to-digital converter, the gizmo that samples audio and converts it into digital data.
If you’re not using off-the-shelf samples, the first step is to record or sample some material. For a more detailed discussion of digital sampling see our Quick Guide To Digital Audio but briefly, there two main sampling considerations – sample rate and sample resolution. The sample rate is how many times the source material is read or sampled per second. The higher the rate, the more accurate the sample. Audio CDs use a sample rate of 44.1kHz. Many samplers offer rates up to 96kHz and some even go as high as 192kHz.
The sample resolution is the measuring scale used to store the sample readings. It’s measured in bits and early samplers had a resolution of 8- or 12-bits. You can calculate the resolution by raising 2 to the power of the number of bits. So, an 8-bit system is 256 (2^8) and a 12-bit system is 4096 (2^12). Each sample in an 8-bit system, for example, must take a value from 0 to 255.
Home on the range
Given the vast dynamic range of natural sounds you can see that 256 values is not going to be terribly accurate. 4096 is better but the 65536 values of 16 bits is better still and this is the resolution used by CDs. Many samplers now offer 24-bit resolution which comes as close to capturing all the nuances of natural sound as you can get. Until they move to 32-bit sampling, that is…
So why not simply record at the highest possible sample rate and resolution? One day we will, but the higher the rate and resolution the more processing power and storage space is required so at the moment we balance what we would like against what we have and what we can afford.
Not everyone has access to a symphony orchestra, exotic percussion and the Vienna Boy’s Choir from which to create their own samples so it’s good to know that dozens of sample producers do and there are thousands of pre-recorded samples you can load into your sampler. Unfortunately, there’s no single standard sample format and samples come in several guises. Most sample producers tend to produce samples in Wave (for the PC) and AIFF (for the Mac) formats and these are as close to a standard as we have at the moment.
Most of the other formats were developed for specific samplers and you might want to consider how important it is for your sampler to be able to read them. In the heyday of the hardware sampler, the Akai sample format was de rigour to read. The Sound Designer II (SDII) format used by Digidesign software is still popular on the Mac, and many collections were created for the SampleCell sampler, originally designed for the Mac but later ported to the PC. Creative Lab’s SoundFont (.sf2) format is popular with hundreds of thousands of SoundBlaster sound card users and is also commonly supported by samplers.
Early samplers stored their samples in RAM as, indeed, do many software samplers. This obviously limits the number and size of the samples you can store and play at any one time. However, increases in computer power have led enterprising companies to develop software samplers such as Nemesys’ GigaSampler and Steinberg’s HALion that can read samples directly from hard disk. At last, size is not an obstacle.
The memory limitations of early samplers meant you could not store samples of any great length and, therefore, you could not hold a note for very long. The solution was the loop. You’d find a central portion of the sample and simply repeat it for as long as the key was held down and then jump to the end of the sample when the key was released.
Finding good loop points was an art, and a difficult one at that. The problem has been alleviated somewhat by samplers’ and computers’ ability to hold more RAM, cheaper RAM prices, the ability to store samples on hard disk, and by better looping facilities in the form of audio editors and dedicated loop-finding software.
Root of the problem
The first stage in preparing a sample for playback is to assign it to a root note. This does not necessarily have to be the pitch at which it plays. This is obvious in the case of drums, for example, but it’s also useful when creating layers and key splits as it means you can play the same pitch from two or more sections of a keyboard.
To play an instrument sample in a realistic manner, you need to be able to play a range of pitches and different volume levels. The simplest way to do this is to assign a single sample to a root note and then pitchshift it to other pitches. However, this makes the note duration shorter or longer than the original sample and the attack phase will be shorter or longer, too. This can be acceptable for pitches say three or four semitones away from the root but you don’t have to transpose a sample very far from its original pitch before it starts to sound unnatural.
The way around this is to use more than one sample, a process known, helpfully, as multisampling.
In an ideal world, each individual note would have its own sample but, as you can imagine, the sampling process would require a lot of time and effort, and the sampler itself would need a lot of resources to store and play all those samples.
But that has not prevented several enterprising companies producing gargantuan sample sets such as Steinberg’s The Grand which includes 1.3Gb of samples!
However, depending on the instrument, you can often get realistic playback using a different sample for only every three or four notes.
But simply having a different sample for each note is not always realistic enough. With most acoustic instruments, the timbre of a note changes with its volume. Louder notes tend to have more high harmonics and often a faster attack time. So we could go another mile and sample each note at several volume levels. On playback, the MIDI velocity determines which sample is played.
This process is known as velocity switching and is set up by selecting velocity points so notes with a velocity below, say, 80 will trigger one particular sample and notes with velocities above 80 will trigger another. This feature works particularly well with drums because their tone changes quite noticeably the harder they are hit harder.
You can do wild and wacky things with velocity switching such as assigning totally different samples to each velocity value. Try it with drum samples so that each time you press a key with a different pressure you’ll get a different drum sound.
Velocity crossfading uses a feature found in many samplers called reverse sensitivity or reverse velocity which reverses the way velocity normally works so the softer you play, the louder the output.
To use this feature, you’d set up a string sample, for example, with reverse sensitivity and combine it with a piano sample. When you play with average pressure you’ll hear both piano and strings, when you play hard you’ll hear mainly piano and when you play soft you’ll hear mainly strings.
Layering and stacking
Layering or stacking is simply assigning different samples to the same key to create mega combination sounds. Of course, you can combine this with velocity tricks, too, so different velocity levels play different sets of samples.
Key splitting and zones
Samplers have many features to help you set up and organise samples. One is the creation of zones, sometimes also know as key splitting. It simply involves dividing the keyboard into several sections or zones, say into octaves, and assigning a different set of samples to each zone. This is useful in performance as it enables you to play several sounds from one keyboard but it’s not such an essential feature if samples are being triggered on a computer via MIDI. However, many samplers use the zone concept to help with the organisation of layers and the setting of velocity levels.
As well as splitting the key range into discrete zones, many samplers allow you to overlap the zones. You could split a keyboard into three sections; the lower playing strings, the upper playing piano and the middle section playing both piano and strings.
Once upon a time if you had all the above features in your sampler you’d be a very happy bunny. Nowadays, however, most samplers have many extras such as built-in filters and effects, LFOs and envelopes. In fact, many have essentially all the features of a full-blown synthesiser except the sound is generated by samples rather than oscillators.
But the main strength of a sampler remains its ability to convincingly reproduce acoustic instruments and natural sounds, and the flexibility of its programming makes it an invaluable instrument in the studio.
For more info…
Zicweb list of software samplers
Music Machine hardware samplers
Yahoo list of sampling resources