funasr_wss_server.py 当python funasr_wss_client.py使用--audio_in方式时，没有正确返回asr，且造成客户端卡死

[wxfred]

🐛 Bug

runtime/python/websocket/funasr_wss_server.py，这个代码好像有bug，funasr_wss_client.py 通过 --audio_in 上传文件这个用法，如果是 -mode offline 直接卡死没有输出；如果是 -mode online或2pass，音频的最后几个字一直识别不出来然后卡死

To Reproduce

1. python funasr_wss_server.py --port 10095 启动服务器
2. python funasr_wss_client.py --host 127.0.0.1 --port 10095 --mode offline --audio_in xxx

Expected behavior

funasr_wss_client.py --audio_in使用任何mode都应该正确返回完成的asr识别内容，且识别完后应该自动关闭

[wxfred]

让ai分析了一下，它给funasr_wss_server.py加了一段这个代码就好了，但是不知道其它场景对不对

                if "audio_fs" in messagejson:
                    websocket.audio_fs = _safe_int(messagejson["audio_fs"], 16000)

                # 新增代码
                if not websocket.is_speaking:
                    if frames_asr and websocket.mode != "online":
                        asyncio.create_task(async_asr(websocket, b"".join(frames_asr)))
                        frames_asr = []
                        frames_asr_online = []
                        websocket.status_dict_asr_online["cache"] = {}
                    elif frames_asr_online:
                        # online: 发剩余帧做最终增量
                        asyncio.create_task(async_asr_online(websocket, b"".join(frames_asr_online)))
                        frames_asr_online = []
                        websocket.status_dict_asr_online["cache"] = {}

                continue

[wxfred]

另外还有个问题，客户端文件传输模式下，代码居然是等10s结束。
服务端和客户端可以新增一个is_end信号，客户端所有文件发送完后，发送一个is_end信号，服务器收到后返回is_end表示全处理完了

funasr_wss_server.py

                if "audio_fs" in messagejson:
                    websocket.audio_fs = _safe_int(messagejson["audio_fs"], 16000)

                # 新增代码
                if not websocket.is_speaking:
                    if frames_asr and websocket.mode != "online":
                        await async_asr(websocket, b"".join(frames_asr))
                        frames_asr = []
                        frames_asr_online = []
                        websocket.status_dict_asr_online["cache"] = {}
                    elif frames_asr_online:
                        await async_asr_online(websocket, b"".join(frames_asr_online))
                        frames_asr_online = []
                        websocket.status_dict_asr_online["cache"] = {}

                if messagejson.get("is_end"):
                    await websocket.send(json.dumps({
                        "mode": websocket.mode,
                        "is_end": True,
                    }))
                # -----------------------

                continue

funasr_wss_client.py

async def record_from_scp(chunk_begin, chunk_size):
    ...
    
    # 新增一个 is_end
    await websocket.send(json.dumps({"is_end": True}, ensure_ascii=False))

    # if not args.mode == "offline":
    #     await asyncio.sleep(2)

    # if args.mode == "offline":
    #     global offline_msg_done
    #     while not offline_msg_done:
    #         await asyncio.sleep(1)

    # await asyncio.sleep(10)

    # await websocket.close()

async def message(id, writer: MeetingWriter):
    ...
            if "mode" not in meg:
                continue

            # 新增
            if meg.get("is_end"):
                await websocket.close()
                break

[wxfred]

客户端这里处理wav文件用 pydub 兼容性更强

        if wav_path.endswith(".pcm"):
            with open(wav_path, "rb") as f:
                audio_bytes = f.read()
        else:
            from pydub import AudioSegment
            audio = AudioSegment.from_file(wav_path)
            audio = audio.set_channels(1).set_frame_rate(sample_rate).set_sample_width(2)
            audio_bytes = audio.raw_data