This area is to share code for generating a basic English language phenome synthesizer. It's an inherently wiki-able project, since it can be viewed as a large number of similar, small modules with a few key characteteristics in common.
It is targeted at Open Music Labs' midiVamp, details of which can not be released at this time.
Parameters needed to begin:
Wavetable size in bytes: ?
Wavetable sample depth: (8 bits ?)
Target pitch: (100 - 250 Hz) ?
Maximum Frequency Content: ?
There should be about 60 (+/- 30) or so phenomes. They should be grouped in a few ways.
- 1. Similar to synthesize, e.g. vowels
2. Similar in sound, e.g. 'p' and 'b'. People who study these for a living divide them into specific catergories with actual names
Here is one proposed grouping:
Vowels: a, e, i, o, u
Consonsants b, d, f, g, j, k, l, m, n, p, s, t, v, z
Other: h, r, wh, y,
Subtractive Synthesis Model
In this model, all vowel sounds begin as short wavetables which can be synthesized from e.g. a C program or octave script. In theory, many of the vowels can use the same fundamental waveform, with a number of different filters applied to it. In theory, many of the consonants can be synthesized based on a white/pink noise waveform and filtered. Nearly every single phenome will utilize one or more envelope waveforms to shape the filters, volume levels, etc.
Taking together, that means there are several common components which we know we will need to synthesize:
- Vowel fundamental
- White noise synthesizer
- Envelope Generator