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
feat: add funasr-server CLI command for one-line API deployment
游雁97ec3ae2ee56
diff --git a/funasr/bin/_server_app.py b/funasr/bin/_server_app.pynew file mode 100644index 000000000..dde92f0bf--- /dev/null+++ b/funasr/bin/_server_app.py@@ -0,0 +1,104 @@+"""Internal: FastAPI app for funasr-server command."""++import tempfile+import time+import os+import re+import logging+from typing import Optional++from fastapi import FastAPI, UploadFile, File, Form, HTTPException+from fastapi.responses import JSONResponse++logger = logging.getLogger("funasr.server")++MODEL_CONFIGS = {+ "sensevoice": {+ "model": "iic/SenseVoiceSmall",+ "vad_model": "fsmn-vad",+ "vad_kwargs": {"max_single_segment_time": 30000},+ },+ "paraformer": {+ "model": "paraformer-zh",+ "vad_model": "fsmn-vad",+ "punc_model": "ct-punc",+ },+ "paraformer-en": {+ "model": "paraformer-en",+ "vad_model": "fsmn-vad",+ },+ "fun-asr-nano": {+ "model": "FunAudioLLM/Fun-ASR-Nano-2512",+ "hub": "hf",+ "trust_remote_code": True,+ "vad_model": "fsmn-vad",+ "vad_kwargs": {"max_single_segment_time": 30000},+ },+}+++def create_app(device: str = "cuda", preload_model: str = "sensevoice") -> FastAPI:+ app = FastAPI(title="FunASR Server", version="1.3.2")+ app.state.models = {}+ app.state.device = device++ def _load_model(name: str):+ if name in app.state.models:+ return app.state.models[name]+ if name not in MODEL_CONFIGS:+ raise ValueError(f"Unknown model: {name}. Available: {list(MODEL_CONFIGS.keys())}")+ from funasr import AutoModel+ cfg = MODEL_CONFIGS[name].copy()+ cfg["device"] = app.state.device+ cfg["disable_update"] = True+ logger.info(f"Loading '{name}' on {device}...")+ t0 = time.time()+ model = AutoModel(**cfg)+ logger.info(f"'{name}' ready in {time.time()-t0:.1f}s")+ app.state.models[name] = model+ return model++ # Pre-load+ _load_model(preload_model)++ @app.post("/v1/audio/transcriptions")+ async def transcribe(+ file: UploadFile = File(...),+ model: str = Form(default="sensevoice"),+ language: Optional[str] = Form(default=None),+ response_format: Optional[str] = Form(default="json"),+ ):+ if model not in MODEL_CONFIGS:+ raise HTTPException(400, f"Unknown model '{model}'. Available: {list(MODEL_CONFIGS.keys())}")+ suffix = os.path.splitext(file.filename)[1] if file.filename else ".wav"+ with tempfile.NamedTemporaryFile(delete=False, suffix=suffix) as tmp:+ tmp.write(await file.read())+ tmp_path = tmp.name+ try:+ asr = _load_model(model)+ kwargs = {"input": tmp_path, "batch_size": 1}+ if language:+ kwargs["language"] = language+ result = asr.generate(**kwargs)+ text = re.sub(r'<\|[^|]*\|>', '', result[0]["text"]).strip()+ if response_format == "verbose_json":+ segments = []+ if "sentence_info" in result[0]:+ for s in result[0]["sentence_info"]:+ segments.append({"start": s.get("start",0)/1000, "end": s.get("end",0)/1000, "text": re.sub(r'<\|[^|]*\|>','',s.get("text","")).strip(), "speaker": s.get("spk")})+ return JSONResponse({"text": text, "segments": segments, "model": model})+ return JSONResponse({"text": text})+ except Exception as e:+ raise HTTPException(500, str(e))+ finally:+ os.unlink(tmp_path)++ @app.get("/v1/models")+ async def list_models():+ return JSONResponse({"object": "list", "data": [{"id": n, "object": "model", "owned_by": "funasr"} for n in MODEL_CONFIGS]})++ @app.get("/health")+ async def health():+ return {"status": "ok", "device": app.state.device, "models_loaded": list(app.state.models.keys())}++ return appdiff --git a/funasr/bin/server.py b/funasr/bin/server.pynew file mode 100644index 000000000..7a4e6dd86--- /dev/null+++ b/funasr/bin/server.py@@ -0,0 +1,66 @@+"""+FunASR Server — OpenAI-compatible speech recognition API.++Usage:+ funasr-server # default: sensevoice on cuda:0, port 8000+ funasr-server --device cpu --port 9000+ funasr-server --model paraformer+"""++import argparse+import sys+++def main():+ parser = argparse.ArgumentParser(+ description="FunASR OpenAI-Compatible API Server",+ formatter_class=argparse.RawDescriptionHelpFormatter,+ epilog="""+Examples:+ funasr-server # Start with SenseVoice on GPU+ funasr-server --device cpu # Start on CPU+ funasr-server --model paraformer # Use Paraformer model+ funasr-server --port 9000 # Custom port++Then use with OpenAI SDK:+ from openai import OpenAI+ client = OpenAI(base_url="http://localhost:8000/v1", api_key="x")+ result = client.audio.transcriptions.create(model="sensevoice", file=open("a.wav","rb"))+""",+ )+ parser.add_argument("--host", default="0.0.0.0", help="Bind address (default: 0.0.0.0)")+ parser.add_argument("--port", type=int, default=8000, help="Port (default: 8000)")+ parser.add_argument("--device", default="cuda", help="Device: cuda, cpu, mps (default: cuda)")+ parser.add_argument("--model", default="sensevoice", help="Pre-load model: sensevoice, paraformer, fun-asr-nano (default: sensevoice)")+ args = parser.parse_args()++ try:+ import uvicorn+ import fastapi+ except ImportError:+ print("Error: funasr-server requires additional packages.")+ print("Install with: pip install fastapi uvicorn python-multipart")+ sys.exit(1)++ # Import and configure the app+ import os+ sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..', '..', 'examples', 'openai_api'))++ # Use inline app to avoid path issues+ from funasr.bin._server_app import create_app++ app = create_app(device=args.device, preload_model=args.model)++ print(f"╔══════════════════════════════════════════════╗")+ print(f"║ FunASR Server v1.3.2 ║")+ print(f"║ Device: {args.device:<8} ║")+ print(f"║ Model: {args.model:<12} ║")+ print(f"║ URL: http://{args.host}:{args.port}/v1 ║")+ print(f"║ Docs: http://{args.host}:{args.port}/docs ║")+ print(f"╚══════════════════════════════════════════════╝")++ uvicorn.run(app, host=args.host, port=args.port)+++if __name__ == "__main__":+ main()diff --git a/setup.py b/setup.pyindex 6c2a81254..2e7545dce 100644--- a/setup.py+++ b/setup.py@@ -154,6 +154,7 @@ setup(entry_points={"console_scripts": ["funasr = funasr.bin.inference:main_hydra",+ "funasr-server = funasr.bin.server:main","funasr-train = funasr.bin.train:main_hydra","funasr-train-ds = funasr.bin.train_ds:main_hydra","funasr-export = funasr.bin.export:main_hydra",