Skip to content

Instantly share code, notes, and snippets.

@otmb
Last active January 20, 2024 20:34
Show Gist options
  • Save otmb/15f1d0b486d57d0363d126d8a19dc78c to your computer and use it in GitHub Desktop.
Save otmb/15f1d0b486d57d0363d126d8a19dc78c to your computer and use it in GitHub Desktop.
whisper.cpp Server Web. Interface

whisper.cpp Server

https://github.com/ggerganov/whisper.cpp/tree/master/examples/server

Setup

  1. whisper.cpp setup
  2. server build and run
$ make server
$ ./server -m ./models/ggml-large-v3.bin -l ja
  1. voice record server
$ python3 -m http.server

whisper.cpp recode file format

// index.html
const wavBuffer = vad.utils.encodeWAV(arr, 1, 16000, 1, 16)
Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 16000 Hz, 1 channels, s16, 256 kb/s
<meta charset="UTF-8" />
<title>whisper.cpp server</title>
<script src="https://cdn.jsdelivr.net/npm/onnxruntime-web@1.13.1/dist/ort.js"></script>
<script src="https://cdn.jsdelivr.net/npm/@ricky0123/vad/dist/index.browser.js"></script>
<script type="module">
try {
const myvad = await vad.MicVAD.new({
positiveSpeechThreshold: 0.5,
negativeSpeechThreshold: 0.5 - 0.15,
preSpeechPadFrames: 2,
redemptionFrames: 8,
frameSamples: 1536,
minSpeechFrames: 3,
onSpeechEnd: async (arr) => {
const wavBuffer = vad.utils.encodeWAV(arr, 1, 16000, 1, 16)
var file = new File([wavBuffer], `file${Date.now()}.wav`)
let formData = new FormData()
formData.append("file", file)
formData.append("response_format", "json")
try {
const resp = await fetch("http://127.0.0.1:8080/inference", {
method: "POST",
body: formData,
})
const resp2 = await resp.json()
console.log(resp2.text)
const result_text = resp2.text;
const text = document.body;
text.setAttribute("data-text", result_text)
} catch (err) {
console.log(err)
}
},
})
myvad.start()
} catch (e) {
console.error("Failed:", e)
}
</script>
<style>
@import url('https://fonts.googleapis.com/css2?family=M+PLUS+Rounded+1c:wght@900&display=swap');
body {
margin: 0;
padding: 0;
border: 0;
display: flex;
flex-wrap: wrap;
justify-content: center;
align-items: flex-end;
align-content: flex-end;
overflow: hidden;
margin: 5vh 5vw;
font-family: 'M PLUS Rounded 1c', sans-serif;
font-size: 10vh;
line-height: 1.1em;
text-align: center;
position: relative;
}
body::before {
content: attr(data-text);
position: absolute;
color: #FF0;
-webkit-text-stroke: 0.5vh #000;
z-index: 30;
}
body::after {
content: attr(data-text);
position: absolute;
color: #FFF;
-webkit-text-stroke: 1.0vh #FFF;
z-index: 20;
}
body {
background-color: lime;
}
</style>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment