Qwen

/

Qwen3-ASR-1.7B

like 885

Follow

Qwen 88.6k

Qwen3-ASR works seamlessly with FunASR toolkit for production deployment:

from funasr import AutoModel

model = AutoModel(
    model="Qwen/Qwen3-ASR-1.7B",
    hub="hf",
    trust_remote_code=True,
    vad_model="funasr/fsmn-vad",
    device="cuda",
)
result = model.generate(input="audio.wav", batch_size=1)
print(result[0]["text"])

FunASR adds:

• VAD — auto-segments long audio
• Speaker diarization — add spk_model="funasr/campplus"
• OpenAI-compatible API — funasr-server --device cuda
• Streaming — WebSocket real-time

Install: pip install funasr>=1.3.3
GitHub: https://github.com/modelscope/FunASR