family misc · 3 variants · unknown
| benchmark | category | best score |
|---|---|---|
| WER | speech-asr | 17.2 |
| WER Earnings-22 | speech-asr | 13.17 |
| WER GigaSpeech | speech-asr | 9.67 |
| WER VoxPopuli | speech-asr | 8.01 |
| WER LibriSpeech | speech-asr | 5.51 |
| WER SPGISpeech | speech-asr | 3.8 |
| WER TED-LIUM | speech-asr | 2.57 |
| variant | kind | quant | bits | vRAM | dl | org | year |
|---|---|---|---|---|---|---|---|
| VibeVoice-ASR-HF | original | bf16 | 16.0 | 198,817 | microsoft | 2026 | |
| VibeVoice-ASR-HF-NF4 | bnb-4bit | 4bit | 539 | Dubedo | 2026 | ||
| VibeVoice-ASR-HF-INT8 | bnb-4bit | 4bit | 216 | Dubedo | 2026 |