The vocal synthesis used is a simple 3 filter setup, one low pass and two band pass. A waveform or noise is passed through the filter bank, depending on whether the phoneme is voiced or unvoiced. The filter values for each phoneme were based on analysis of spectrograms of my own voice. 18 different voice types are available, and a "shift" multiplier value can be applied to the set of formants. Every word is rendered as a set of envelopes for all of the parameters. Each individual phoneme sound is selected for every word or phrase, together with its pitch, amplitude and duration values. These words or phrases can be saved and recalled individually, to allow a "dictionary" to be built up.
<< back to sequencer