Qwen3-ASR-0.6B-hf

Home/Models/Qwen/Qwen3-ASR-0.6B-hf

Qwen/Qwen3-ASR-0.6B-hf

kind: original
quant: bf16 (16-bit)
family: qwen
modality: automatic-speech-recognition
root model: Qwen/Qwen3-ASR-0.6B-hf
provider: Qwen
license: apache-2.0
language: zh, en, yue, ar, de, fr, es, pt, id, it, ko, ru
context: —
library: transformers
vRAM (GB): —
size on disk: —
downloads: 156,789
likes: 43
created: —
last modified: 2026-07-22T11:58:47+00:00 (2026)
huggingface: https://huggingface.co/Qwen/Qwen3-ASR-0.6B-hf
tags: transformerssafetensorsqwen3_asrautomatic-speech-recognitionzhenyueardefresptiditkoruthvijatrhimsnlsv

benchmark	category	score	conf	evidence
WER Earnings-22	speech-asr	10.72	high	`\| Qwen3-ASR-0.6B-hf \| 6.31 \| 10.57 \| 10.72 \| 7.65 \| 1.69 \| 3.97 \| 2.74 \| 6.80 \|`
WER AMI	speech-asr	10.57	high	`\| Qwen3-ASR-0.6B-hf \| 6.31 \| 10.57 \| 10.72 \| 7.65 \| 1.69 \| 3.97 \| 2.74 \| 6.80 \|`
WER GigaSpeech	speech-asr	7.65	high	`\| Qwen3-ASR-0.6B-hf \| 6.31 \| 10.57 \| 10.72 \| 7.65 \| 1.69 \| 3.97 \| 2.74 \| 6.80 \|`
WER VoxPopuli	speech-asr	6.8	high	`\| Qwen3-ASR-0.6B-hf \| 6.31 \| 10.57 \| 10.72 \| 7.65 \| 1.69 \| 3.97 \| 2.74 \| 6.80 \|`
WER	speech-asr	6.31	high	`\| Qwen3-ASR-0.6B-hf \| 6.31 \| 10.57 \| 10.72 \| 7.65 \| 1.69 \| 3.97 \| 2.74 \| 6.80 \|`
WER SPGISpeech	speech-asr	2.74	high	`\| Qwen3-ASR-0.6B-hf \| 6.31 \| 10.57 \| 10.72 \| 7.65 \| 1.69 \| 3.97 \| 2.74 \| 6.80 \|`

none scraped.