Phoneme Synthesis

This phase converts the phonemes in a phoneme sequence to audio. This can be done in several ways:

1. Calling out to an external phoneme synthesizer such as MBROLA;
2. Using Spectral or Formant parameters such as in the Klatt or eSpeak synthesizers;
3. Using a diphone database and concatenating the waveforms such as in the Cepstral or MBROLA synthesizers.

Each of the supported phoneme synthesizers expose a set of voices that are described using RDF metadata.