【Hackathon 8th No.5】在 PaddleSpeech 中复现 DAC 模型

[DrRyanHuang]

PR types

New features

PR changes

Models

Describe

• 【Hackathon 8th】开源贡献个人挑战赛（尝鲜版） Paddle#70746

需要先下载数据集

train/build_dataset.folders

part	path
speech_fb	- 🚧 /data/daps/train (https://ccrma.stanford.edu/~gautham/Site/daps.html)
speech_hq	- ✅ /data/vctk (pdspeech已支持) - 🎯 /data/vocalset(https://zenodo.org/records/1203819) - 🚧 /data/read_speech (DNS-Challenge) - 🚧 /data/french_speech (DNS-Challenge)
speech_uq	- 🚧 /data/emotional_speech/ (DNS-Challenge) - /data/common_voice/ (https://commonvoice.mozilla.org/en/datasets) - 🚧 /data/german_speech/ (DNS-Challenge) - 🚧 /data/russian_speech/ (DNS-Challenge) - 🚧 /data/spanish_speech/ (DNS-Challenge)
music_hq	- /data/musdb/train (https://zenodo.org/records/1117372)
music_uq	- /data/jamendo (https://mtg.github.io/mtg-jamendo-dataset/)
general	- /data/audioset/data/unbalanced_train_segments/ - /data/audioset/data/balanced_train_segments/

---

val/build_dataset.folders
speech_hq:
- /data/daps/val
music_hq:
- /data/musdb/test
general:
- /data/audioset/data/eval_segments/

---

test/build_dataset.folders
speech_hq:
- /data/daps/test
music_hq:
- /data/musdb/test
general:
- /data/audioset/data/eval_segments/