ASR Websocket方式是否支持mp3音频

PaddlePaddle / PaddleSpeech

Easy-to-use Speech Toolkit including Self-Supervised Learning model, SOTA/Streaming ASR with punctuation, Streaming TTS with text frontend, Speaker Verification System, End-to-End Speech Translation and Keyword Spotting. Won NAACL2022 Best Demo Award.

Apache License 2.0

11.12k stars 1.85k forks source link

使用ASR Websocket方式（ws://{server}:{port}/paddlespeech/asr/streaming）传入MP3音频数据返回一两次result:“”后websocket直接断开，{status:1006, reason: ""}。但是发送{ "name": "test.mp3", "signal": "start", "nbest": 1 } 又是返回 {status: 'ok', signal: 'server_ready' }。同样方式，换成 .wav，没有问题。所以是不支持 mp3，还是bug？

相关关键代码如下（uniapp小程序）

this.recorder.start({
    duration: 20000,
    numberOfChannels: 1,
    format: 'mp3',
    frameSize: 5, // 乱取的
    sampleRate: 16000
})

rd.onFrameRecorded(res => {
    console.log('recorder frame', res)
    if (!this.asrUseHttp) {
        const { isLastFrame, frameBuffer } = res
        this.wsSendData(this.asrWs,
            frameBuffer,
            isLastFrame
        )
    }
})

this.wsSendData(this.asrWs, {
    name: '123.mp3',
    signal: 'end',
    nbset: 1
})

this.wsSendData(this.asrWs, {
        task: '123.mp3',
    signal: 'start',
    nbset: 1,
})

PaddlePaddle / PaddleSpeech

ASR Websocket方式是否支持mp3音频 #3714