Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
docs: fix vLLM version requirement, add AutoModelVLLM usage to vllm_inference.md
xiaoyunchong.xyc55c035ff4d17
diff --git a/examples/industrial_data_pretraining/fun_asr_nano/docs/vllm_inference.md b/examples/industrial_data_pretraining/fun_asr_nano/docs/vllm_inference.mdindex c3b39ec44..f4860d648 100644--- a/examples/industrial_data_pretraining/fun_asr_nano/docs/vllm_inference.md+++ b/examples/industrial_data_pretraining/fun_asr_nano/docs/vllm_inference.md@@ -15,7 +15,7 @@ This architecture allows vLLM to efficiently batch multiple requests and leverag```bash# Core dependenciespip install funasr>=1.3.0-pip install vllm>=0.11.0+pip install vllm>=0.12.0pip install safetensorspip install tiktoken@@ -25,7 +25,23 @@ pip install modelscope## Quick Start-### Python API+### AutoModelVLLM (Recommended)++```python+from funasr.auto.auto_model_vllm import AutoModelVLLM++# Generic interface - auto-detects model type, works for all LLM-based ASR models+model = AutoModelVLLM(+ model="FunAudioLLM/Fun-ASR-Nano-2512",+ hub="ms",+ tensor_parallel_size=1,+)++results = model.generate(["audio.wav"], language="中文")+print(results[0]["text"])+```++### FunASRNanoVLLM (Direct API)```pythonfrom funasr.models.fun_asr_nano.inference_vllm import FunASRNanoVLLM