Name		Name	Last commit message	Last commit date
apex		apex
common		common
data_preprocessing		data_preprocessing
debug_epoch		debug_epoch
fastpitch		fastpitch
filelists		filelists
hifigan		hifigan
hiftnet		hiftnet
langid		langid
original_voice_samples		original_voice_samples
scripts		scripts
triton		triton
waveglow		waveglow
.gitignore		.gitignore
README.md		README.md
app.py		app.py
gradio_gui.py		gradio_gui.py
gradio_gui_katri.py		gradio_gui_katri.py
gradio_gui_uralic.py		gradio_gui_uralic.py
models.py		models.py
original_voice_samples.zip		original_voice_samples.zip
prepare_dataset.py		prepare_dataset.py
run_training_cluster_s.sh		run_training_cluster_s.sh
symbols.py		symbols.py
syn_hifigan.py		syn_hifigan.py
syn_hiftnet.py		syn_hiftnet.py
syn_k_univnet_multi.py		syn_k_univnet_multi.py
syn_k_univnet_multi_quant.py		syn_k_univnet_multi_quant.py
syn_k_univnet_ural.py		syn_k_univnet_ural.py
train_1_with_plot_multilang.py		train_1_with_plot_multilang.py

Repository files navigation

FastPitchMulti

Experimental multi-lingual FastPitch

What's done:

Conditioning on language and speaker labels
Dataset and preprocessing of Sámi data
Combined character set for the Sámi languages
Train a model on Sámi languages
Selecting Estonian data
Processing Estonian data
Train a model on Sámi x 3, Finnish, Estonian

Ideas:

Move the language embedding to the very beginning of the encoder

About

Experimental multi-lingual FastPitch

0 stars

2 watching

Report repository

Releases

Packages

Used by

Contributors

Languages