Modelpublic

bezzam/tiny-processor-qwen3asr

sourceHugging Faceapache-2.0updated 3mo agoView on Hugging Face

0likes9downloads

Model Card

Qwen3-ASR (Transformers native)

Overview

The Qwen3-ASR family includes Qwen3-ASR-1.7B and Qwen3-ASR-0.6B, which support language identification and ASR for 52 languages and dialects. Both leverage large-scale speech training data and the strong audio understanding capability of their foundation model, Qwen3-Omni. The 1.7B version achieves state-of-the-art performance among open-source ASR models and is competitive with the strongest proprietary commercial APIs.

Key features:

—All-in-one: Supports language identification and speech recognition for 30 languages and 22 Chinese dialects, including English accents from multiple countries and regions.
—Excellent and Fast: High-quality and robust recognition under complex acoustic environments. Qwen3-ASR-0.6B reaches 2000× throughput at a concurrency of 128. Both models support streaming/offline unified inference with a single model and handle long audio.
—Forced Alignment: Qwen3-ForcedAligner-0.6B supports timestamp prediction for arbitrary units within up to 5 minutes of speech in 11 languages, surpassing E2E-based forced-alignment models in accuracy.

Model Architecture

Available Checkpoints

Model	Supported Languages	Supported Dialects	Inference Mode	Audio Types
Qwen/Qwen3-ASR-1.7B-hf & Qwen/Qwen3-ASR-0.6B-hf	Chinese (zh), English (en), Cantonese (yue), Arabic (ar), German (de), French (fr), Spanish (es), Portuguese (pt), Indonesian (id), Italian (it), Korean (ko), Russian (ru), Thai (th), Vietnamese (vi), Japanese (ja), Turkish (tr), Hindi (hi), Malay (ms), Dutch (nl), Swedish (sv), Danish (da), Finnish (fi), Polish (pl), Czech (cs), Filipino (fil), Persian (fa), Greek (el), Hungarian (hu), Macedonian (mk), Romanian (ro)	Anhui, Dongbei, Fujian, Gansu, Guizhou, Hebei, Henan, Hubei, Hunan, Jiangxi, Ningxia, Shandong, Shaanxi, Shanxi, Sichuan, Tianjin, Yunnan, Zhejiang, Cantonese (HK), Cantonese (Guangdong), Wu, Minnan	Offline / Streaming	Speech, Singing Voice, Songs with BGM
Qwen/Qwen3-ForcedAligner-0.6B-hf	Chinese, English, Cantonese, French, German, Italian, Japanese, Korean, Portuguese, Russian, Spanish	—	NAR	Speech

Usage

Qwen3-ASR is supported natively in 🤗 Transformers. Until it is part of an official Transformers release, install from source:

bash

pip install git+https://github.com/huggingface/transformers

Simple transcription

apply_transcription_request handles chat-template formatting for you and is the recommended entry point.

python

from transformers import AutoProcessor, AutoModelForMultimodalLM

model_id = "Qwen/Qwen3-ASR-0.6B-hf"
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForMultimodalLM.from_pretrained(model_id, device_map="auto")
print(f"Model loaded on {model.device} with dtype {model.dtype}")

inputs = processor.apply_transcription_request(
    audio="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-ASR-Repo/asr_en.wav",
).to(model.device, model.dtype)

output_ids = model.generate(**inputs, max_new_tokens=256)
generated_ids = output_ids[:, inputs["input_ids"].shape[1]:]

# Raw output includes language tag and <asr_text> marker
raw = processor.decode(generated_ids)[0]
print(f"Raw: {raw}")

# Parsed output: dict with "language" and "transcription"
parsed = processor.decode(generated_ids, return_format="parsed")[0]
print(f"Parsed: {parsed}")

# Extract only the transcription text
transcription = processor.decode(generated_ids, return_format="transcription_only")[0]
print(f"Transcription: {transcription}")

"""
Raw: language English<asr_text>Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel.
Parsed: {'language': 'English', 'transcription': 'Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel.'}
Transcription: Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel.
"""

Language hint

Pass a language hint to skip auto-detection.

python

from transformers import AutoProcessor, AutoModelForMultimodalLM

model_id = "Qwen/Qwen3-ASR-0.6B-hf"
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForMultimodalLM.from_pretrained(model_id, device_map="auto")

# Without language hint (auto-detect)
inputs = processor.apply_transcription_request(
    audio="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-ASR-Repo/asr_zh.wav",
).to(model.device, model.dtype)
output_ids = model.generate(**inputs, max_new_tokens=256)
generated_ids = output_ids[:, inputs["input_ids"].shape[1]:]
print(f"Auto-detect: {processor.decode(generated_ids, return_format='transcription_only')[0]}")

# With language hint (language code or full name both accepted)
inputs = processor.apply_transcription_request(
    audio="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-ASR-Repo/asr_zh.wav",
    language="Chinese",  # or "zh"
).to(model.device, model.dtype)
output_ids = model.generate(**inputs, max_new_tokens=256)
generated_ids = output_ids[:, inputs["input_ids"].shape[1]:]
print(f"With hint:   {processor.decode(generated_ids, return_format='transcription_only')[0]}")

Batch inference

Pass a list of audio paths and optional languages to transcribe multiple files in one call.

python

from transformers import AutoProcessor, AutoModelForMultimodalLM

model_id = "Qwen/Qwen3-ASR-0.6B-hf"
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForMultimodalLM.from_pretrained(model_id, device_map="auto")

audio = [
    "https://huggingface.co/datasets/bezzam/audio_samples/resolve/main/librispeech_mr_quilter.wav",
    "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-ASR-Repo/asr_zh.wav",
]

inputs = processor.apply_transcription_request(
    audio, language=[None, "zh"],
).to(model.device, model.dtype)

output_ids = model.generate(**inputs, max_new_tokens=256)
generated_ids = output_ids[:, inputs["input_ids"].shape[1]:]
transcriptions = processor.decode(generated_ids, return_format="transcription_only")

for i, text in enumerate(transcriptions):
    print(f"Audio {i + 1}: {text}")

Chat template

apply_transcription_request is a convenience wrapper around apply_chat_template. Use the chat template directly for more control, such as providing a language hint via a system message.

python

from transformers import AutoProcessor, Qwen3ASRForConditionalGeneration

model_id = "Qwen/Qwen3-ASR-0.6B-hf"
processor = AutoProcessor.from_pretrained(model_id)
model = Qwen3ASRForConditionalGeneration.from_pretrained(model_id, device_map="auto")

chat_template = [
    [
        {"role": "system", "content": [{"type": "text", "text": "English"}]},
        {
            "role": "user",
            "content": [
                {
                    "type": "audio",
                    "path": "https://huggingface.co/datasets/bezzam/audio_samples/resolve/main/librispeech_mr_quilter.wav",
                },
            ],
        },
    ],
    [
        {
            "role": "user",
            "content": [
                {
                    "type": "audio",
                    "path": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-ASR-Repo/asr_zh.wav",
                },
            ],
        },
    ],
]

inputs = processor.apply_chat_template(
    chat_template, tokenize=True, return_dict=True,
).to(model.device, model.dtype)

output_ids = model.generate(**inputs, max_new_tokens=256)
generated_ids = output_ids[:, inputs["input_ids"].shape[1]:]
transcriptions = processor.decode(generated_ids, return_format="transcription_only")
for text in transcriptions:
    print(text)

Training / Fine-tuning

python

from transformers import AutoProcessor, Qwen3ASRForConditionalGeneration

model_id = "Qwen/Qwen3-ASR-0.6B-hf"
processor = AutoProcessor.from_pretrained(model_id)
model = Qwen3ASRForConditionalGeneration.from_pretrained(model_id, device_map="auto")
model.train()

chat_template = [
    [
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel.",
                },
                {
                    "type": "audio",
                    "path": "https://huggingface.co/datasets/bezzam/audio_samples/resolve/main/librispeech_mr_quilter.wav",
                },
            ],
        }
    ],
]

inputs = processor.apply_chat_template(
    chat_template, tokenize=True, return_dict=True, output_labels=True,
).to(model.device, model.dtype)

loss = model(**inputs).loss
print("Loss:", loss.item())
loss.backward()

Forced alignment (word-level timestamping)

Use Qwen3ASRForTokenClassification to obtain word-level timestamps from a transcript. Transcribe first with the ASR model, then align with the forced aligner.

Supported languages: Chinese, English, Cantonese, French, German, Italian, Japanese, Korean, Portuguese, Russian, Spanish.

Japanese requires nagisa and Korean requires soynlp: pip install nagisa soynlp

python

import torch
from transformers import AutoProcessor, AutoModelForMultimodalLM, AutoModelForTokenClassification

asr_model_id = "Qwen/Qwen3-ASR-0.6B-hf"
aligner_model_id = "Qwen/Qwen3-ForcedAligner-0.6B-hf"

asr_processor = AutoProcessor.from_pretrained(asr_model_id)
asr_model = AutoModelForMultimodalLM.from_pretrained(asr_model_id, device_map="auto")

aligner_processor = AutoProcessor.from_pretrained(aligner_model_id)
aligner_model = AutoModelForTokenClassification.from_pretrained(
    aligner_model_id, dtype=torch.bfloat16, device_map="auto"
)

audio_url = "https://huggingface.co/datasets/bezzam/audio_samples/resolve/main/librispeech_mr_quilter.wav"

# Step 1: Transcribe
inputs = asr_processor.apply_transcription_request(audio=audio_url)
inputs = inputs.to(asr_model.device, asr_model.dtype)
output_ids = asr_model.generate(**inputs, max_new_tokens=256)
generated_ids = output_ids[:, inputs["input_ids"].shape[1]:]
parsed = asr_processor.decode(generated_ids, return_format="parsed")[0]
transcript = parsed["transcription"]
language = parsed["language"] or "English"

# Step 2: Prepare alignment inputs
aligner_inputs, word_lists = aligner_processor.prepare_forced_aligner_inputs(
    audio=audio_url, transcript=transcript, language=language,
)
aligner_inputs = aligner_inputs.to(aligner_model.device, aligner_model.dtype)

# Step 3: Run forced aligner
with torch.inference_mode():
    outputs = aligner_model(**aligner_inputs)

# Step 4: Decode timestamps
timestamps = aligner_processor.decode_forced_alignment(
    logits=outputs.logits,
    input_ids=aligner_inputs["input_ids"],
    word_lists=word_lists,
    timestamp_token_id=aligner_model.config.timestamp_token_id,
)[0]

for item in timestamps:
    print(f"{item['text']:<20} {item['start_time']:>8.3f}s → {item['end_time']:>8.3f}s")

"""
Word                  Start (s)    End (s)
------------------------------------------
Mr                        0.560      0.800
Quilter                   0.800      1.280
is                        1.280      1.440
the                       1.440      1.520
apostle                   1.520      2.080
...
"""

Pipeline usage

python

from transformers import pipeline

model_id = "Qwen/Qwen3-ASR-0.6B-hf"
pipe = pipeline("any-to-any", model=model_id, device_map="auto")

chat_template = [
    {
        "role": "user",
        "content": [
            {
                "type": "audio",
                "path": "https://huggingface.co/datasets/bezzam/audio_samples/resolve/main/librispeech_mr_quilter.wav",
            },
        ],
    }
]
outputs = pipe(text=chat_template, return_full_text=False)
raw_text = outputs[0]["generated_text"]

# Use processor helper to extract transcription
transcription = pipe.processor.extract_transcription(raw_text)
print(f"Transcription: {transcription}")

Speed & Memory Improvements

Torch compile

Both the ASR and forced aligner models support torch.compile. The forced aligner is a particularly good fit because it runs a single forward pass with no autoregressive decoding, making it ideal for bulk timestamping workflows.

On an A100 we observed ~2.5× speed-up for the forced aligner and ~2.4× for ASR generate at batch size 4.

python

import torch
from transformers import AutoProcessor, AutoModelForMultimodalLM

model_id = "Qwen/Qwen3-ASR-0.6B-hf"
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForMultimodalLM.from_pretrained(model_id, dtype=torch.bfloat16).to("cuda").eval()

audio_url = "https://huggingface.co/datasets/bezzam/audio_samples/resolve/main/librispeech_mr_quilter.wav"
inputs = processor.apply_transcription_request(
    audio=[audio_url] * 4,
).to("cuda", torch.bfloat16)

model.forward = torch.compile(model.forward)

# Warmup
with torch.inference_mode():
    for _ in range(3):
        _ = model.generate(**inputs, max_new_tokens=256, do_sample=False)

# Inference
with torch.inference_mode():
    output_ids = model.generate(**inputs, max_new_tokens=256, do_sample=False)

generated_ids = output_ids[:, inputs["input_ids"].shape[1]:]
print(processor.decode(generated_ids, return_format="transcription_only")[0])

Evaluation

WER on the HuggingFace Open ASR Leaderboard (26 June 2026):

Model	Mean WER	AMI	Earnings22	GigaSpeech	LS Clean	LS Other	SPGISpeech	VoxPopuli
Qwen3-ASR-1.7B-hf	5.59	9.26	9.88	7.25	1.24	2.92	2.58	5.99
Qwen3-ASR-0.6B-hf	6.31	10.57	10.72	7.65	1.69	3.97	2.74	6.80

Citation

bibtex

@article{Qwen3-ASR,
  title={Qwen3-ASR Technical Report},
  author={Xian Shi, Xiong Wang, Zhifang Guo, Yongqi Wang, Pei Zhang, Xinyu Zhang, Zishan Guo,
          Hongkun Hao, Yu Xi, Baosong Yang, Jin Xu, Jingren Zhou, Junyang Lin},
  journal={arXiv preprint arXiv:2601.21337},
  year={2026}
}