Sampled-Sounds and Synthesized Sounds

What Are They, And How Do They Work?

Other than the sound of a bell tone, which is a simple sine-wave, with no overtones, the distinctive sound of each musical instrument is recognized by its overtones, which are tittle squiggly waves superimposed on a sine-wave, as in the lower graph.


Sine Wave Graph – a bell or clear tone might look like this


Overtones Graph – a musical instrument sound would look like this

Synthesized Sounds

It is possible, using the output of multiple oscillators (electronic components that generate a sine-wave), running at different frequencies (how many oscillations per second), to combine the output of those oscillators to produce wave-forms that look like the distinctive wave-forms of all of the musical instruments.

These wave-forms may not be exact, but they can be close enough that most listeners will not know the difference between the generated wave-form, and the actual instrument sound.

Generating musical sounds by this method, is called synthesis, or synthesized sounds.

When synthesizers were first invented, this was how the musical instrument sounds were generated. As the technology improved, the instrument sounds generated this way, became more and more like the instrument sounds they were trying to mimic.

In the 1980's, the Roland D20 was an example of state-of-the-art sound synthesis.


A Roland D20 Synthesizer

Here is an example of music improvised on the KeyMusician Keyboard, played using the Roland D20 synthesizer's sound generator:

Synthesized Sound Example - Roland D20

Not all synthesized sounds are intended to mimic the sounds of traditional musical instruments. The same sound synthesis techniques can yield amazing, and unique sounds that don't sound like any particular musical instrument. The ZynAddSubFX synthesizer has many sounds that fall into this category.


The Main Window of the ZynAddSubFX Synthesizer

Here is a composition of mine, called “The Zen Of Zyn”, played on the KeyMusician Keyboard, using the ZynAddSubFX synthesizer:

Synthesized Sound Example - ZynAddSubFX

Sampled Sounds

In the 1980's, when I bought my Roland D20, another type of synthesizer was beginning to take hold. These new synthesizers used sampled sound. At present, most synthesizers (and VST instruments) use sampled sound, and the quality of that sound has increased to an amazing level.

But even before then, there was a synthesizer using tape-loops of recorded sound, called the Mellotron, made famous by bands such as The Moody Blues, or Emerson, Lake, & Palmer.

So just what is sampled sound?

The audio files people listen to, whether from a CD, or downloaded MP3 file, are the result of continuous sampling of the sound of music being performed. The sound waves of the music, produce the variations in air-pressure, and are picked up by a microphone.

What sampling is, is a continuous stream of measurements of the positive (or negative) pressure of the air where music is being played, taken at a regular, but very small interval. For example, the music in a WAV file (as would come on an audio CD), is a continuous set of air-pressure measurements, taken 44100 times per second, with each measurement stored as a 16-bit floating-point (number with a decimal point) number.

MP3 files are similarly a set of samplings of the air-pressure, except there Is a way of compressing successive equal-value samples into a count of those equal-value samples, and the value of that repeated samples. There may be other methods of compressing the data as well, defined in the MP3 method.

So regardless whether the music was recorded from authentic instruments or not, the sound of that recording, when it comes to us, is actually sampled sound.

Making Notes From Sampled Sound

To put things very simply, the sound of each note that can be played by an instrument, or sung by a singer, is recorded, and saved in a collection of sounds. Depending on the note you select (by pushing a key), the proper recorded note sound is retrieved, and played.

But how does that work, where not all notes are the same length?

It turns out, that the sound of each note can be broken down into three parts: the Attack sound, the Sustain-sound, and the Release-sound. These parts of the sound of a note are referred to by the acronym ASR, meaning Attack, Sustain, and Release.

Note: There is another often mentioned part of the sound, coming after the Attack, called Decay, where the initially loud attack sound goes down to a lower volume level.

To illustrate the parts of a note sound, think of the sound of the meow of a cat.

The attack portion of this sound is “me”, the sustain part of the sound is a sort of short-A sound, like “aaaaaa”, which stays the same for most of the meow sound. At the end of that variable-length sustain sound, is the release sound, which is what happens when the cat stops making the meow sound, namely the “w' part of the sound.

If you played a note made from a sampled cat's “meow”, the key you pressed would select the pitch (frequency), and the attack sound (“me”) would immediately play. After that, the sustain part of the sound (“aaaa”) will continue playing for as long as you hold down the key. When you release the key, the release part of the sound (“w”) would play.

The attack part of the sound, and the release part of the sound are constant length, and the sustain part of the sound is a variable-length, depending on how long you hold the note's key down. The length of the attack part of the sound, and the release part of the sound, are not necessarily the same length.

Vibrato

Though not present in our cat's-meow sample, other really expressive things can be a part of the sustain part of the sound. One of these is called vibrato, which is a very expressive, wavering of the sustained sound.

For sampled instrument sounds not including vibrato, you can use the Modulation control (the default control of the “Assigned” slider on the KMK performance panes), to generate an artificial vibrato in the sound, oscillating at a low (4 cycles per second?) rate.

If the sustained part of the sound includes vibrato (as it does using the FluidR3_GM.sf2 soundfont), adding different vibrato using the Modulation control can produce a not-so-good sound, so beware of that. Some soundfonts work fine using the modulation control, others do not. This problem may be specific to certain instrument sounds.

Soundfonts

The collection of instrument note sounds, are often stored in a file called a soundfont (SF2 file), though there are a variety of file formats for storing sampled sound.

There is software available for making your own soundfonts, using the sound you record of notes being played (or sung).

The actual recorded sound within a soundfont, is a sequence of 16-bit floating-point samples for each note.

VST Instruments

Newer VST instruments using sampled sounds use 32-bit floating-point values for the samples, and the sound does sound better than what you would get from a soundfont (having only 16-bit floating-point samples). Some even use 64-bit floating-point samples.

Though I can perceive an improvement of the sound using 32-bit samples rather than 16-bit samples, I personally don't perceive any difference between using 64-bit samples over 32-bit samples. But some people may be able to tell a difference.

There are other ways of improving the quality of the generated sound, such as sampling the air-pressure at a higher rate of speed, such as 48000 times per second, as opposed to the 44100 times per second used on audio CDs.

Though these higher resolution sampling methods can produce a noticeably good effect in a performance, if that performance is recorded, and produced as an audio file, chances are it will be reduced back to 16-bit samples, so take that into consideration. Also, depending on the sound system you use, or the earphones, you might not be able to hear the better sound.


Main window of Ableton Live – a high-end VST host, with sampled-sound VST Instruments, showing various controls

Nevertheless, here is a composition of mine, called “Intermezzo # 5”, played on the KeyMusician Keyboard, using Ableton Live, which is a high-end VST host, with high-end VST instruments, using sampled sound (with at least 32-bit samples). The recording (alas) uses 16-bit samples:

VST Instrument Sampled Sound Example

So to some who wonder, if sampled sounds are real, or are they fake? I respond with a hearty “Oh, they're very real!”

I hope this explanation of the technology you're using with the KeyMusician Keyboard is useful, and interesting, and that it was enjoyable.

- Aere

Index Of All Newsletter Articles