Update README.md
Browse files
README.md
CHANGED
|
@@ -4,4 +4,17 @@ datasets:
|
|
| 4 |
- mozilla-foundation/common_voice_16_1
|
| 5 |
size_categories:
|
| 6 |
- 1K<n<10K
|
| 7 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
- mozilla-foundation/common_voice_16_1
|
| 5 |
size_categories:
|
| 6 |
- 1K<n<10K
|
| 7 |
+
---
|
| 8 |
+
## Don't attempt to identify the identity of the speaker in the voices.
|
| 9 |
+
I agree to download original **public domain** dataset.
|
| 10 |
+
|
| 11 |
+
## Audio Info
|
| 12 |
+
- 22050hz mono wav - for modern tts (original are 32khz mono mp3)
|
| 13 |
+
- vadded 250ms (little bit long)
|
| 14 |
+
- Size-class:Huge(9522 files) - it's take long time to train,but possible to create high-quality tts from scratch.
|
| 15 |
+
- Volume-level:OK - no need to gain.
|
| 16 |
+
- Noise-level:OK - no need to denoise
|
| 17 |
+
## Detail
|
| 18 |
+
- part of [common_voice_16_1 ja](https://huggingface.co/datasets/mozilla-foundation/common_voice_16_1)
|
| 19 |
+
- audios are vadded with [silero-vad](https://github.com/snakers4/silero-vad)
|
| 20 |
+
- text are cleaned simply by openjtalk.(this mean possible contain miss reading and miss converting)
|