MS-Tacotron2
Tacotron2 based multi-speaker text to speech
Reference
Model
Three places adding speaker embedding layers
Data preparation
data
:dictionaries for different language
dataset
:where put training data and test data
Train step
python -m multiproc.py train.py -o outdir15/ -l logdir -c ./outdir5/tacotron2_statedict.pt --warm_start
Experiments
E1 (Jan-19)
location: 115 outdir14 data:
E2 (Jan-24)
location: 114 outdir15 data:
Install
TO DO LIST