A typical Midi synthesizer can deal with up to 16 independent channels of notes and controllers, and each of these channels can use a difference voice (also known in Midi as a patch).

Voices

A voice is the sound of a particular instrument.  Some examples are: Grand Piano, Harpsichord, Trumpet, Acoustic Guitar, Electric Guitar, Bass Guitar, Cello, Violin, Marimba, Flute, etc.

Depending on the quality of the synthesizer, these can sound very realistic when used effectively with appropriate controllers (Pitch Bend, Modulation, etc).

Synthesizers usually offer (at least) 128 different voices, which are arranged as Midi values 0 to 127.  These are often renumbered for display purposes as voices 1 to 128, simply by adding 1.  Songtrix offers a choice.

While the Midi Specification does not dictate exactly how each voice should sound, it does state (in a standard called General Midi) what type of voice should be found at each voice number.

This is so that, for example a piano sound in a Midi file produced on one system will still generate a piano sound when played on another system, even if the piano sound itself is different.

Voice Envelopes

There are two major characteristics of voices which you should consider when making your selection.  There is the tone of a voice, or what it sounds like (piano, guitar, trumpet, etc).

Secondly there is the envelope of the voice, which is the dynamic shape of the sound level as the note is turned on and off.  For example, a piano has a sharp, percussive beginning, then fades even while the key is pressed, while a tuba has a gradual onset, and remains at full level until the key is released.

Envelopes are often described in terms of Attack, Decay, Sustain and Release.  Attack is how sharply the sound begins, followed by a decay to steady level.  This level sustains until the sound is turned off, when it is released and fades.

The diagram above is typical for a tuba or other brass instrument sound, with a slow attack, mild decay, constant sustain and slow release.  Compare with the envelope for a piano sound, shown below.

The piano sound has a sharp attack and decay, a fading sustain, and a rapid release as the dampers stop the sound.

The envelope will have an impact on your selection of voice.  If the part is playing quick, short notes, you want a rapid attack and release so each note can be heard clearly.  If the part is playing long, slow notes, you want a constant sustain so the notes will not fade while being held on.

Voice Banks

Some synthesizers are capable of producing many more than 128 different voices.  To allow Midi files to access these extra voices, they are arranged into banks, or sets of 128 voices.

The first 128 voices are accessed with just the voice number (or with bank 0/voice 0 through to bank 0/voice 127).  Then there might be another 128 voices from bank 1/voice 0 through to bank 1/voice 127.

Since there can be 16384 voice banks, there can theoretically be over 2 million voices available.  Of course, no synthesizer has anywhere near that many, and the unused banks and voice numbers simply have no effect.

Often voices are not arranged neatly into full banks but are scattered, presumably for internal technical reasons, in patchy clusters across various partial banks (except for bank 0, which has the General Midi voices). 

In these cases, Songtrix can use instrument definition files to identify exactly where voices are located, so you don't need to waste time searching for them.

Track Setup

In Songtrix, each track can be assigned its own voice and bank using the track setup controls.