candle-whisper / index.html
lmz's picture
minor ui change (#2)
e521042
raw
history blame
11.2 kB
<html>
<head>
<meta content="text/html;charset=utf-8" http-equiv="Content-Type" />
<title>Candle Whisper Rust/WASM</title>
</head>
<body></body>
</html>
<!doctype html>
<html>
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<style>
@import url("https://fonts.googleapis.com/css2?family=Source+Code+Pro:wght@200;300;400&family=Source+Sans+3:wght@100;200;300;400;500;600;700;800;900&display=swap");
html,
body {
font-family: "Source Sans 3", sans-serif;
}
</style>
<script src="https://cdn.tailwindcss.com"></script>
<script type="module">
// base url for audio examples
const AUDIO_BASE_URL =
"https://huggingface.co/datasets/Narsil/candle-examples/resolve/main/";
// models base url
const MODELS = {
tiny_en: {
base_url:
"https://huggingface.co/openai/whisper-tiny.en/resolve/refs%2Fpr%2F17/",
},
};
const whisperWorker = new Worker("./whisperWorker.js", {
type: "module",
});
async function classifyAudio(
weightsURL, // URL to the weights file
modelID, // model ID
tokenizerURL, // URL to the tokenizer file
mel_filtersURL, // URL to the mel filters file
audioURL, // URL to the audio file
updateStatus // function to update the status
) {
return new Promise((resolve, reject) => {
whisperWorker.postMessage({
weightsURL,
modelID,
tokenizerURL,
mel_filtersURL,
audioURL,
});
whisperWorker.addEventListener("message", (event) => {
console.log(event.data);
if ("status" in event.data) {
updateStatus(event.data);
}
if ("error" in event.data) {
reject(new Error(event.data.error));
}
if (event.data.status === "complete") {
resolve(event.data);
}
});
});
}
// keep track of the audio URL
let audioURL = null;
function setAudio(src) {
const audio = document.querySelector("#audio");
audio.src = src;
audio.controls = true;
audio.hidden = false;
document.querySelector("#detect").disabled = false;
audioURL = src;
}
// add event listener to audio buttons
document.querySelectorAll("#audios-select > button").forEach((target) => {
target.addEventListener("click", (e) => {
const value = target.dataset.value;
const href = AUDIO_BASE_URL + value;
setAudio(href);
});
});
//add event listener to file input
document.querySelector("#file-upload").addEventListener("change", (e) => {
const target = e.target;
if (target.files.length > 0) {
const href = URL.createObjectURL(target.files[0]);
setAudio(href);
}
});
// add event listener to drop-area
const dropArea = document.querySelector("#drop-area");
dropArea.addEventListener("dragenter", (e) => {
e.preventDefault();
dropArea.classList.add("border-blue-700");
});
dropArea.addEventListener("dragleave", (e) => {
e.preventDefault();
dropArea.classList.remove("border-blue-700");
});
dropArea.addEventListener("dragover", (e) => {
e.preventDefault();
dropArea.classList.add("border-blue-700");
});
dropArea.addEventListener("drop", (e) => {
e.preventDefault();
dropArea.classList.remove("border-blue-700");
const url = e.dataTransfer.getData("text/uri-list");
const files = e.dataTransfer.files;
if (files.length > 0) {
const href = URL.createObjectURL(files[0]);
setAudio(href);
} else if (url) {
setAudio(url);
}
});
// add event listener to detect button
document.querySelector("#detect").addEventListener("click", async () => {
if (audioURL === null) {
return;
}
const modelID = document.querySelector("#model").value;
const modelURL = MODELS[modelID].base_url + "model.safetensors";
const tokenizerURL = MODELS[modelID].base_url + "tokenizer.json";
classifyAudio(
modelURL,
modelID,
tokenizerURL,
"mel_filters.safetensors",
audioURL,
updateStatus
)
.then((result) => {
console.log("RESULT", result);
const { output } = result;
const text = output.map((segment) => segment.dr.text).join(" ");
console.log(text);
document.querySelector("#output-status").hidden = true;
document.querySelector("#output-generation").hidden = false;
document.querySelector("#output-generation").textContent = text;
})
.catch((error) => {
console.error(error);
});
});
function updateStatus(data) {
const { status, message } = data;
const button = document.querySelector("#detect");
if (status === "decoding" || status === "loading") {
button.disabled = true;
button.textContent = message;
} else if (status === "complete") {
button.disabled = false;
button.textContent = "Transcribe Audio";
}
}
</script>
</head>
<body class="container max-w-4xl mx-auto p-4">
<main class="grid grid-cols-1 gap-8 relative">
<span class="absolute text-5xl -ml-[1em]"> 🕯️ </span>
<div>
<h1 class="text-5xl font-bold">Candle Whisper</h1>
<h2 class="text-2xl font-bold">Rust/WASM Demo</h2>
<p class="max-w-lg">
Transcribe audio in the browser using rust/wasm with an audio file.
This demo uses the
<a
href="https://huggingface.co/openai/"
target="_blank"
class="underline hover:text-blue-500 hover:no-underline"
>
OpenAI Whisper models
</a>
and WASM runtime built with
<a
href="https://github.com/huggingface/candle/"
target="_blank"
class="underline hover:text-blue-500 hover:no-underline"
>Candle
</a>
</p>
</div>
<div>
<label for="model" class="font-medium">Models Options: </label>
<select
id="model"
class="border-2 border-gray-500 rounded-md font-light"
>
<option value="tiny_en" selected>tiny.en (151 MB)</option>
</select>
</div>
<!-- drag and drop area -->
<div class="relative">
<div
id="drop-area"
class="flex flex-col items-center justify-center border-2 border-gray-300 border-dashed rounded-xl relative h-48 w-full overflow-hidden"
>
<div
class="flex flex-col items-center justify-center space-y-1 text-center"
>
<svg
width="25"
height="25"
viewBox="0 0 25 25"
fill="none"
xmlns="http://www.w3.org/2000/svg"
>
<path
d="M3.5 24.3a3 3 0 0 1-1.9-.8c-.5-.5-.8-1.2-.8-1.9V2.9c0-.7.3-1.3.8-1.9.6-.5 1.2-.7 2-.7h18.6c.7 0 1.3.2 1.9.7.5.6.7 1.2.7 2v18.6c0 .7-.2 1.4-.7 1.9a3 3 0 0 1-2 .8H3.6Zm0-2.7h18.7V2.9H3.5v18.7Zm2.7-2.7h13.3c.3 0 .5 0 .6-.3v-.7l-3.7-5a.6.6 0 0 0-.6-.2c-.2 0-.4 0-.5.3l-3.5 4.6-2.4-3.3a.6.6 0 0 0-.6-.3c-.2 0-.4.1-.5.3l-2.7 3.6c-.1.2-.2.4 0 .7.1.2.3.3.6.3Z"
fill="#000"
/>
</svg>
<div class="flex text-sm text-gray-600">
<label
for="file-upload"
class="relative cursor-pointer bg-white rounded-md font-medium text-blue-950 hover:text-blue-700"
>
<span>Drag and drop your audio here</span>
<span class="block text-xs">or</span>
<span class="block text-xs">Click to upload</span>
</label>
</div>
<input
id="file-upload"
name="file-upload"
type="file"
accept="audio/*"
class="sr-only"
/>
</div>
<audio
id="audio"
hidden
controls
class="w-full p-2 select-none"
></audio>
</div>
</div>
<div>
<div class="flex flex-wrap gap-3 items-center" id="audios-select">
<h3 class="font-medium">Examples:</h3>
<button
data-value="samples_jfk.wav"
class="text-gray-500 border border-gray-500 rounded-md p-2 underline hover:no-underline"
>
<span>jfk.wav</span>
<span class="text-xs block"> (352 kB)</span>
</button>
<button
data-value="samples_a13.wav"
class="text-gray-500 border border-gray-500 rounded-md p-2 underline hover:no-underline"
>
<span>a13.wav</span>
<span class="text-xs block"> (960 kB)</span>
</button>
<button
data-value="samples_mm0.wav"
class="text-gray-500 border border-gray-500 rounded-md p-2 underline hover:no-underline"
>
<span>mm0.wav</span>
<span class="text-xs block new"> (957 kB)</span>
</button>
<button
data-value="samples_gb0.wav"
class="text-gray-500 border border-gray-500 rounded-md p-2 underline hover:no-underline"
>
<span>gb0.wav </span>
<span class="text-xs block">(4.08 MB)</span>
</button>
<button
data-value="samples_gb1.wav"
class="text-gray-500 border border-gray-500 rounded-md p-2 underline hover:no-underline"
>
<span>gb1.wav </span>
<span class="text-xs block">(6.36 MB)</span>
</button>
<button
data-value="samples_hp0.wav"
class="text-gray-500 border border-gray-500 rounded-md p-2 underline hover:no-underline"
>
<span>hp0.wav </span>
<span class="text-xs block">(8.75 MB)</span>
</button>
</div>
</div>
<div>
<button
id="detect"
disabled
class="bg-gray-700 hover:bg-gray-800 text-white font-normal py-2 px-4 rounded disabled:bg-gray-300 disabled:cursor-not-allowed"
>
Transcribe Audio
</button>
</div>
<div>
<h3 class="font-medium">Transcription:</h3>
<div
class="min-h-[250px] bg-slate-100 text-gray-500 p-4 rounded-md flex flex-col gap-2"
>
<p hidden id="output-generation" class="grid-rows-2"></p>
<span id="output-status" class="m-auto font-light"
>No transcription results yet</span
>
</div>
</div>
</main>
</body>
</html>