ParallelWaveGAN

What's Changed

• Add LibriTTS-R recipe by @ShigekiKarita in https://github.com/kan-bayashi/ParallelWaveGAN/pull/414
• Update version and add ci for newest pytorch by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/418

New Contributors

• @ShigekiKarita made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/414

Full Changelog: https://github.com/kan-bayashi/ParallelWaveGAN/compare/v0.6.0...v.0.6.1

What's Changed

• temporary fixed protobuf version by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/366
• Fix #367 by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/369
• CVSS_C Recipe by @ftshijt in https://github.com/kan-bayashi/ParallelWaveGAN/pull/386
• support VQ-VAE and discrete symbol models by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/387
• use isort by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/389
• Apply new black ✨ 🍰 ✨ by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/390
• Hubert-discrete symbol-based HiFiGAN with duration predictor by @ftshijt in https://github.com/kan-bayashi/ParallelWaveGAN/pull/388
• Fixes to discrete-symbol-based vocoder by @ftshijt in https://github.com/kan-bayashi/ParallelWaveGAN/pull/394
• fix ci by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/398
• add singing vocoder and update f0 & excitation input feats by @DrWelles in https://github.com/kan-bayashi/ParallelWaveGAN/pull/391
• fix norm issue by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/409
• fix pyreaper installation by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/411
• fix the case when segments has different #lines from wav.scp by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/413

New Contributors

• @DrWelles made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/391

Full Changelog: https://github.com/kan-bayashi/ParallelWaveGAN/compare/v0.5.5...v0.6.0

What's Changed

• add recipe for kiritan & ofuton_p_utagoe db (singing voice synthesis) by @PeterGuoRuc in https://github.com/kan-bayashi/ParallelWaveGAN/pull/330
• Add recipe for Opencpop by @ftshijt in https://github.com/kan-bayashi/ParallelWaveGAN/pull/332
• add causal option for HiFiGAN by @chomeyama in https://github.com/kan-bayashi/ParallelWaveGAN/pull/326
• Fix HiFiGAN compatibility by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/334
• add recipe for natsume (singing voice synthesis) by @PeterGuoRuc in https://github.com/kan-bayashi/ParallelWaveGAN/pull/336
• Update readme with pre-trained models on svs and demonstration by @ftshijt in https://github.com/kan-bayashi/ParallelWaveGAN/pull/335
• Add recipes and pretrained models for CSD (Korean&English) and KiSIng (Mandarin) databases by @ftshijt in https://github.com/kan-bayashi/ParallelWaveGAN/pull/342
• Add new recipe PJS (singing voice synthesis) by @A-Quarter-Mile in https://github.com/kan-bayashi/ParallelWaveGAN/pull/347
• add no7singing training by @frankxu2004 in https://github.com/kan-bayashi/ParallelWaveGAN/pull/349
• Add icelandic by @G-Thor in https://github.com/kan-bayashi/ParallelWaveGAN/pull/354
• add tag_or_url for download_pretrained_model by @roholazandie in https://github.com/kan-bayashi/ParallelWaveGAN/pull/361
• Apply black by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/362
• Update to v0.5.5 by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/363

New Contributors

• @PeterGuoRuc made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/330
• @chomeyama made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/326
• @A-Quarter-Mile made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/347
• @frankxu2004 made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/349
• @G-Thor made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/354
• @roholazandie made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/361

Full Changelog: https://github.com/kan-bayashi/ParallelWaveGAN/compare/v0.5.4...v0.5.5

What's Changed

• add kss recipe by @windtoker in https://github.com/kan-bayashi/ParallelWaveGAN/pull/306
• Fix a noise shape of StyleMelGANGenerator to export ONNX model by @c-bata in https://github.com/kan-bayashi/ParallelWaveGAN/pull/312
• add recipe for oniku_kurumi_utagoe db (singing voice synthesis) by @ftshijt in https://github.com/kan-bayashi/ParallelWaveGAN/pull/323
• update documentation and correct the download link for oniku-db by @ftshijt in https://github.com/kan-bayashi/ParallelWaveGAN/pull/324
• Fix an error librosa update by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/327
• Add pytorch 1.10.x CI by @kan-bayashi in https://github.com/kan-bayashi/ParallelWaveGAN/pull/328

New Contributors

• @windtoker made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/306
• @c-bata made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/312
• @ftshijt made their first contribution in https://github.com/kan-bayashi/ParallelWaveGAN/pull/323

Full Changelog: https://github.com/kan-bayashi/ParallelWaveGAN/compare/v0.5.3...v0.5.4

• Added pretrained models.

• Added pretrained models
• Fixed typo
• Introduced filelock to prevent from duplicated download

• Support normalize-before option in decoding
• Support log_base option in preprocessing
• Update pre-trained models

• Support HiFiGAN
• Support StyleMelGAN
• Refactored

• Support Pytorch 1.7
• Add JSSS recipe
• Support the case where wav.scp and segments have a different number of lines

• Minor fix of pretrained model save directory

• Simplify decoding part
• Add load_model function
• Add inference method in each genearator
• Add download_pretraiend_model function to download from google drive directly

• Fix PQMF #204

• Fixed wrong loss naming
• Fixed shuffle in distributed training
• Fixed pad mode
• Updated to be able to change PQMF paramters

• Fix numba version issue
• Support multi-key hdf5 scp format
• Support npy scp format

What's new

• Support Multi-band MelGAN
• Support Full-band MelGAN

What's new

Scripts

• Fix the update order #131 #132

Recipes

• Add egs/libritts/voc1
• Update structure #134
• Add template recipe #141

What's new

Scripts

• Support --pretrain option in training
• Support --skip-wav-copy option in normalization
• Support scp style input for all /bin scripts for ESPnet compatibility
• Better parallelization (much faster in the case of the large dataset)

Recipes

• Fix format: npy support
• Add VCTK recipe
• Add melgan.v3 config
• Add parallel_wavegan.v3 config

What's new

• Support more recipes
• Support new residual discriminator
• Support MelGAN generator
• Support MelGAN discriminator
• And more refactoring...

Stable version release.