Synthesizing New Utterances Using Existing Voices

Synthesizing a new set of utterances using HTS-engine

This assumes you already have produced fullcontext .lab files for the utterances you want to synthesize.
  1. Drop your .lab files into data/labels/gen.
  2. Modify data/scp/gen.scp to point to your new gen labels.
  3. Re-run step $ENGIN only.
  4. The new utterances will be in gen/qst001/ver1/hts_engine. Nothing that was already there will get overwritten.

Synthesizing utterances with HTS-engine from the command line

[[TODO]]

Synthesizing a new set of utterances using SPTK

[[TODO]]

Synthesizing a new set of utterances using STRAIGHT

[[TODO]]

Synthesizing utterances from the Festival REPL

[[TODO]] you need to load the voice into Festival

Synthesizing new utterances from a Clustergen voice

Under yourvoicedir, run:

./bin/do_clustergen cg_test tts tts_test etc/txt.done.data

tts_test will be the name of the directory where the synthesized .wav files will go (under test/). txt.done.data is the name of the data file of utterances that you want to synthesize.