Skip to content

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

5.4 KB
Raw Blame
import importlib.util
import sys
import types
from pathlib import Path

import numpy as np


REPO_ROOT = Path(__file__).resolve().parents[1]
SERVICE_PATH = (
    REPO_ROOT
    / "examples"
    / "industrial_data_pretraining"
    / "fun_asr_nano"
    / "serve_vllm.py"
)


def load_service_module(monkeypatch):
    fastapi_stub = types.ModuleType("fastapi")

    class FastAPIStub:
        def __init__(self, *args, **kwargs):
            pass

        def on_event(self, *args, **kwargs):
            return lambda func: func

        def post(self, *args, **kwargs):
            return lambda func: func

        def websocket(self, *args, **kwargs):
            return lambda func: func

    fastapi_stub.FastAPI = FastAPIStub
    fastapi_stub.File = lambda *args, **kwargs: None
    fastapi_stub.Form = lambda *args, **kwargs: None
    fastapi_stub.UploadFile = object
    fastapi_stub.WebSocket = object
    fastapi_stub.WebSocketDisconnect = Exception

    responses_stub = types.ModuleType("fastapi.responses")

    class JSONResponseStub:
        def __init__(self, content=None):
            self.content = content

    responses_stub.JSONResponse = JSONResponseStub

    funasr_stub = types.ModuleType("funasr")
    funasr_stub.AutoModel = object

    nano_stub = types.ModuleType("funasr.models.fun_asr_nano.inference_vllm")
    nano_stub.FunASRNanoVLLM = object

    vad_stub = types.ModuleType("funasr.models.fsmn_vad_streaming.dynamic_vad")
    vad_stub.DynamicStreamingVAD = object

    monkeypatch.setitem(sys.modules, "fastapi", fastapi_stub)
    monkeypatch.setitem(sys.modules, "fastapi.responses", responses_stub)
    monkeypatch.setitem(sys.modules, "uvicorn", types.ModuleType("uvicorn"))
    monkeypatch.setitem(sys.modules, "soundfile", types.ModuleType("soundfile"))
    monkeypatch.setitem(sys.modules, "torch", types.ModuleType("torch"))
    monkeypatch.setitem(sys.modules, "funasr", funasr_stub)
    monkeypatch.setitem(sys.modules, "funasr.models", types.ModuleType("funasr.models"))
    monkeypatch.setitem(
        sys.modules,
        "funasr.models.fun_asr_nano",
        types.ModuleType("funasr.models.fun_asr_nano"),
    )
    monkeypatch.setitem(
        sys.modules, "funasr.models.fun_asr_nano.inference_vllm", nano_stub
    )
    monkeypatch.setitem(
        sys.modules,
        "funasr.models.fsmn_vad_streaming",
        types.ModuleType("funasr.models.fsmn_vad_streaming"),
    )
    monkeypatch.setitem(
        sys.modules, "funasr.models.fsmn_vad_streaming.dynamic_vad", vad_stub
    )

    module_name = "serve_vllm_under_test"
    sys.modules.pop(module_name, None)
    spec = importlib.util.spec_from_file_location(module_name, SERVICE_PATH)
    module = importlib.util.module_from_spec(spec)
    assert spec.loader is not None
    spec.loader.exec_module(module)
    return module


def test_openai_verbose_json_keeps_segment_speaker(monkeypatch):
    module = load_service_module(monkeypatch)

    response = module.build_openai_verbose_json(
        {
            "duration": 1.2,
            "text": "hello",
            "segments": [
                {
                    "start": 0.0,
                    "end": 1.2,
                    "text": "hello",
                    "words": [{"word": "hello", "start": 0.0, "end": 1.2}],
                    "speaker": "SPK0",
                }
            ],
        },
        language="en",
    )

    assert response["segments"][0]["speaker"] == "SPK0"


def test_process_audio_downmixes_stereo_before_resampling(monkeypatch):
    module = load_service_module(monkeypatch)
    captured = {}

    librosa_stub = types.ModuleType("librosa")

    def fake_resample(audio, orig_sr, target_sr):
        assert audio.ndim == 1
        captured["resample_input"] = audio.copy()
        assert orig_sr == 8000
        assert target_sr == 16000
        return np.repeat(audio, 2)

    librosa_stub.resample = fake_resample
    monkeypatch.setitem(sys.modules, "librosa", librosa_stub)

    class EngineStub:
        def generate(self, inputs, **kwargs):
            captured["engine_input"] = inputs[0]
            return [{"text": "ok"}]

    module._engine = EngineStub()
    stereo_audio = np.column_stack(
        [np.zeros(8000, dtype=np.float32), np.ones(8000, dtype=np.float32)]
    )

    result = module.process_audio(
        stereo_audio,
        sr=8000,
        use_vad=False,
        use_spk=False,
        use_timestamp=False,
    )

    assert np.allclose(captured["resample_input"], 0.5)
    assert captured["engine_input"].shape == (16000,)
    assert np.allclose(captured["engine_input"], 0.5)
    assert result["duration"] == 1.0
    assert result["text"] == "ok"


def test_load_engine_limits_vad_segments_to_30_seconds(monkeypatch):
    module = load_service_module(monkeypatch)
    captured = {}

    class EngineFactoryStub:
        @classmethod
        def from_pretrained(cls, **kwargs):
            return object()

    def auto_model_stub(**kwargs):
        captured.update(kwargs)
        return object()

    module.FunASRNanoVLLM = EngineFactoryStub
    module.AutoModel = auto_model_stub
    module._engine = None
    module.load_engine(
        types.SimpleNamespace(
            model="fun-asr-nano",
            hub="ms",
            device="cpu",
            dtype="bf16",
            max_model_len=4096,
            gpu_memory_utilization=0.5,
            vad_model="fsmn-vad",
            spk_model="",
        )
    )

    assert captured["max_single_segment_time"] == 30000