Question

因此,这是一种骗局。

我想写一个网络应用程序,记录用户所说的话,然后将样本发送给服务器一侧处理。

The algorithm I have in mind is as follows:

start a recording session once the user has clicked on a button
wait for the user to have said a single word (assuming he know he s supposed to say a single word)
stop the recording once he said it
send the samples to the server, say using HTTP
process the signal on the server side
send some response back to the user.

There are several solution for Voice Activity Detection in Java,C#, and other high level languages I presume. However, I wish this part would to be done on the Client side (otherwise, I ll have to send too much data from the client to the server which is highly inefficient) I.E in javascript and HTML5.

I m not an experienced web developer, so my questions are: Is this feasible? Is there a library for that (I haven t found any)? What would be the best approach in approaching the problem?

Answer 1

这里有一只冰 v,有浏览器支持:

https://www.npmjs.com/ Package/@ricky0123/vad。

You will need to add the CDNs:

<script src="https://cdn.jsdelivr.net/npm/onnxruntime-web/dist/ort.js"></script>
<script src="https://cdn.jsdelivr.net/npm/@ricky0123/vad/dist/index.browser.js"></script>
<script>
async function startVad(){
      const myvad = await vad.MicVAD.new({
        onFrameProcessed: (probabilities) => {

        },
        onSpeechStart: () => {},
        onVADMisfire: () => {},
        onSpeechEnd: (audio) => {console.log("spoke");},
      })
      myvad.start()

}
startVad();
</script>

友情链接