Keyboard players have always been called on to fill in the sound of missing instruments. The trend accelerated in the 1960s thanks to the Mellotron, which played tape recordings of actual instruments such as strings and flute when you pressed the keys. In the ’80s, digital samplers gave musicians a radical new ability: Any sound at all could be triggered from a MIDI keyboard.
Samplers operate by playing back samples — digital recordings of sounds. The samples are in computer-type RAM memory, which loses its contents when the power goes off, so they have to be stored on a computer disk when the sampler is turned off. USB memory sticks are sometimes used instead of disk storage.
Many samplers today exist in the form of computer software. Hardware instruments such as the Yamaha Motif and Korg Triton series include sampling among their other features, but if you’re looking for a dedicated sampler you’ll find that computer-based sampling offers some potent advantages. The big screen is handy for editing, the computer probably has lots of RAM already (RAM tends to be an optional add-on with hardware instruments), and you’ll be able to run your sampler as a plug-in instrument (usually VST or AudioUnits format) within a host multitrack recording progam.
SAMPLING IN A NUTSHELL
Explaining the features found in today’s software samplers would take many pages. Briefly, samplers operate by playing back previously recorded sounds, such as a trumpet note, a cymbal crash, a spoken vocal phrase, or a drum loop. A single preset in a sampler may use only one sample, which is transposed up or down depending on which key you play. Or a preset may use dozens of samples, which are assigned to different keys or key zones. A sampled grand piano, for instance, might use a different recording for each of the 88 keys, plus have several different samples per key to respond to playing with different velocities. A group of samples like this is called a multisample.
Creating good-sounding multisamples is a lot of work. Most software samplers ship with a variety of high-quality multisamples that are ready to play. You can edit these if you like — for example, by changing the filter cutoff frequency so that the sound is brighter or more muted. You can also record your own samples and assign them to the keyboard. A few software samplers are set up to record new samples themselves, but many of them require that you use a separate piece of software to do the actual recording. Once you’ve recorded a sample and stored it on your computer’s hard drive, you’ll be able to load it into the sampler.
Samplers have the ability to turn a short sound into a long one by looping it (see Figure 1 above). When a sample is looped, the loop will keep playing for as long as you hold down a key (or otherwise maintain a MIDI note-on message). Some software samplers also have the ability to stream long samples directly from the computer’s hard drive. A sample streamed from disk can easily be too large to fit into your computer’s RAM. Disk streaming lets the sampler play sounds such as the natural decay of a piano’s notes that last for many seconds and can’t be looped well because they’re constantly changing. No hardware-based sampler has yet been built that includes streaming playback from a hard drive.
FILE FORMATS
Samplers can load raw audio data. This is usually in the form of WAV or AIFF files; the lower-quality MP3 format tends not to be supported by many samplers. They can also load preset files in their own format. The preset files contain both the audio data itself and information on how it is to be played — which samples are assigned to which keys, settings for envelope generators, and so on.
Many samplers can also load presets in formats that come from other samplers. For instance, a sampler from Company X might be able to load files in Apple EXS24, Steinberg HALion, TASCAM GigaStudio, and other formats. However, there’s no guarantee that the presets will sound exactly the same on your sampler as they did on the original instrument. Each sampler has some features not found on others. While the audio files themselves should play flawlessly, the process of interpreting the data in the preset will inevitably be imperfect.
SLICING BEATS
Because sampled drum loops are such a key ingredient in modern pop music, many samplers have special features for handling beats.
First, you may be able to import (load) audio files in formats that contain both the audio itself and tempo information or “markers” that show where the drum hits are located. The three most popular formats are Apple Loops, REX files (which usually have the file extension .RX2 tacked onto their name), and Acidized WAV. After loading the file, you should be able to speed up or slow down the tempo of your host sequencer and have the beat synchronize with the tempo.
Second, your sampler may have editing features with which you can slice up a recorded beat. You may be able to put markers on the individual hits within the beat, then slice up the sample at the markers and assign the sound within each slice to a separate MIDI key. Some samplers will then export the timing information on the beat as a MIDI clip: When you drag it into a sequencer track, the sequencer can trigger the sliced-up notes in the beat one at a time. This makes the beat almost completely plastic: You can add drum fills, substitute a new snare sound for the old one, and much more.
SAMPLERS VS. EVERYTHING ELSE
The great advantage of sampling is that you can play the actual sound of another instrument — not a synthesized trumpet tone, for instance, but the sound of a real trumpet. The sounds of acoustic instruments are full of subtle details that listeners recognize and appreciate.
The disadvantage of sampling is that these details will sound exactly the same each time the sample is played. Musicians who play acoustic instruments are constantly varying their technique, either consciously or intuitively, to produce expressive musical phrases. As a result, a sampler can easily sound great when playing a single note but stiff and artificial when playing a whole phrase. By using such techniques as multiple articulations (different samples that play back depending on velocity, or can be switched by pressing a pre-assigned key, button, or other controller) and “round-robin” programming (alternating through similar but non-identical samples of the same note to avoid a robotic, “machine-gun” sound), the software engineers who design samplers have made real progress in getting around this problem, but it’s still an issue.
Synthesizers are good at emulating some types of acoustic instruments, but not so good at others. For instance, the FM synthesis technique that the Yamaha DX7 made widely popular in the ’80s can make decent electric piano and marimba sounds, but it’s not so great for bowed strings.
In recent years, synthesis methods using a technique called physical modeling have made strides in producing the sonic details heard in acoustic instruments, but physical modeling software has other limitations. Physical modeling uses lots of your computer’s CPU resources (playing samples requires much less computer power), and only the most popular instruments, such as brass, electric piano, and various classic analog synthesizers, have been modeled. Developing a physical model requires serious expertise, but anyone who has a microphone can record and play a sample! With a sampler, you can turn the pots and pans in your kitchen into a percussion kit.
In sum, a sampler isn’t the right solution for every musical problem, but samplers are an absolutely essential tool in most computer musicians’ kits.