Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -33,7 +33,7 @@ Cleaning steps:
|
|
33 |
* Filtered on samples with >2 upvotes and <1 downvotes]
|
34 |
* Removed non voice audio at start and end through pytorch VAD
|
35 |
|
36 |
-
Tokenization
|
37 |
* Audio tokenized through [EnCodec by Meta](https://github.com/facebookresearch/encodec)
|
38 |
* Using 24khz pre-trained model, and target bandwidth of 1.5
|
39 |
* Represented in text as audio_token_0 - audio_token_1023
|
|
|
33 |
* Filtered on samples with >2 upvotes and <1 downvotes]
|
34 |
* Removed non voice audio at start and end through pytorch VAD
|
35 |
|
36 |
+
Tokenization:
|
37 |
* Audio tokenized through [EnCodec by Meta](https://github.com/facebookresearch/encodec)
|
38 |
* Using 24khz pre-trained model, and target bandwidth of 1.5
|
39 |
* Represented in text as audio_token_0 - audio_token_1023
|