Update README.md
Browse filesAdd link to the blog post.
README.md
CHANGED
|
@@ -16,7 +16,7 @@ We present a voice-safety classification model that can be used for voice-toxici
|
|
| 16 |
The model has been distilled into the [WavLM](https://arxiv.org/abs/2110.13900) architecture from a larger teacher model.
|
| 17 |
All the model training has been conducted with Roblox internal voice chat datasets,
|
| 18 |
using both machine and human-labeled data, with over 100k hours of training data in total.
|
| 19 |
-
We have also published a blog post about this work.
|
| 20 |
|
| 21 |
The model supports eight languages: English, Spanish, German, French, Portuguese, Italian, Korean, and Japanese.
|
| 22 |
It classifies the input audio into six toxicity classes in a multilabel fashion. The class labels are as follows:
|
|
|
|
| 16 |
The model has been distilled into the [WavLM](https://arxiv.org/abs/2110.13900) architecture from a larger teacher model.
|
| 17 |
All the model training has been conducted with Roblox internal voice chat datasets,
|
| 18 |
using both machine and human-labeled data, with over 100k hours of training data in total.
|
| 19 |
+
We have also published [a blog post](https://corp.roblox.com/newsroom/2025/04/updating-ml-voice-safety-more-languages) about this work.
|
| 20 |
|
| 21 |
The model supports eight languages: English, Spanish, German, French, Portuguese, Italian, Korean, and Japanese.
|
| 22 |
It classifies the input audio into six toxicity classes in a multilabel fashion. The class labels are as follows:
|