Dataset · Hugging Face
JayLL13/vivoice-relabeled
Vivoice relabeled using Qwen3-ASR The original dataset: capleaf/viVoice Sample rate: 44.1 kHz [!NOTE]How-tos: I use the Qwen/Qwen3-ASR-1.7B models then update the label for the sample (audio - text) which have WER (word error rate) < 15%.
| Author / organization | JayLL13 |
|---|---|
| Downloads | 1.5K (as reported by Hugging Face) |
| Likes | 3 |
| Created | 7 March 2026 |
| Last updated at source | 21 March 2026 |
Popularity, last 2 days
1,452 now. Counts as reported by the source, snapshotted daily by Silytics.
More from JayLL13
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.