Sample-Based Synthesis

Synthesize Using Samplers

A sample-based synthesis is a form of audio synthesis that mainly involves using samplers, either hardware or software. The main difference between this audio synthesis and the others is that it uses sampled sounds or instruments instead of oscillators with fundamental waveforms such as saw, square, triangle, etc.  

Back In The Day

At the beginning of the second half of the 20th century, before digital sampling was used practically, machines like Mellotron used analog tape decks to playback sampled sounds. Later on, sample-based synthesis ideas evolved with the birth of more powerful samplers in the 1970s and 1980s. The concept behind this form of synthesis was to emulate real instruments. This is done by recalling actual samples of these instruments upon striking the keys on the keyboard. These samplers were expensive back then and could only offer a scarce sample rate and bit depth, often resulting in grainy and aliased sound. Not to mention that the expense of memory limited them and utilized the shortest possible length of the sampled sounds. Moving forward, in recent times, computers have become capable of providing much larger memory and processing speed, so we were introduced to software samplers with many more editing possibilities.  

Main Aspects

So, there are three main issues to address in sample-based synthesis:  

1. Looping

Looping extends the duration of sampled sounds played by a musical keyboard. If the musician holds down a key, the sampler should scan “ seamlessly” through the note until the musician releases the key. Furthermore, this is done by specifying beginning and ending  “loop points“ in the sampled sound. After the attack of the note is finished, the sampler reads repeatedly through the looper part of the wavetable until the key is released. Most of the latest samplers provide automatic methods for finding prospective loop points. One of these methods is to perform pitch detection on the sampled sound. This means that the pitch detection algorithm searches for repeating patterns in the wavetable that indicate a fundamental pitch period. Therefore, the pitch period is the time interval that spans one cycle of a periodic waveform. Once the pitch has been estimated, the sampler suggests a pair of loop points that match the same number of pitch periods in the waveform. This algorithm tends to create smooth loops that are constant in pitch.

2. Pitch Shifting

Indeed, storing every note played by an acoustic instrument in an inexpensive sampler may not be possible. This is because these samplers hold only every 3rd or 4th semitone and obtain intermediate notes by shifting the pitch of a nearby stored note. If you record a sound into a sampler memory and play it back with different keys, the sampler carries out the same pitch-shifting technique. The side effect of the simple pitch shifting is that the sound duration increases or decreases depending on the key pressed. There are two methods of simple pitch shifting to be mentioned.

Method 1:

Varying the clock frequency of the output DAC changes the sampling rate. That changes the pitch up or down and changes the duration.

Method 2:

Sample rate conversion ( resampling the signal in the digital domain ) shifts the pitch inside of the sampler and allows the playback at a constant sampling rate for all pitches.  

3. Data Reduction

A way to reduce data stored in a sampler is to limit sample resolution or quantization. Another way is to lower the sampling rate. This diminishes the number of samples stored per unit of time at the cost of shrinking the audio bandwidth. In addition, a more sophisticated way of data reduction starts from an analysis stage. It stores sound in a data-reduced form, along with control functions that approximately reconstitute it.