Update README.md
#1
by
jpc
- opened
README.md
CHANGED
@@ -1,3 +1,27 @@
|
|
1 |
---
|
2 |
license: mit
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: mit
|
3 |
+
task_categories:
|
4 |
+
- text-to-speech
|
5 |
+
language:
|
6 |
+
- en
|
7 |
+
pretty_name: WhisperSpeech
|
8 |
---
|
9 |
+
|
10 |
+
# The WhisperSpeech Dataset
|
11 |
+
|
12 |
+
This dataset contains data to train SPEAR TTS-like text-to-speech models that utilized semantic tokens derived from the OpenAI Whisper
|
13 |
+
speech recognition model.
|
14 |
+
|
15 |
+
We currently provide semantic and acoustic tokens for the LibriLight and LibriTTS datasets (English only).
|
16 |
+
|
17 |
+
Acoustic tokens:
|
18 |
+
- 24kHz EnCodec 6kbps (8 quantizers)
|
19 |
+
|
20 |
+
Semantic tokens:
|
21 |
+
- Whisper tiny VQ bottleneck trained on a subset of LibriLight
|
22 |
+
|
23 |
+
Available LibriLight subsets:
|
24 |
+
- `small`/`medium`/`large` (following the original dataset division but with `large` excluding the speaker `6454`)
|
25 |
+
- a separate ≈1300hr single-speaker subset based on the `6454` speaker from the `large` subset for training single-speaker TTS models
|
26 |
+
|
27 |
+
We plan to add more acoustic tokens from other codecs in the future.
|