49 lines
1.7 KiB
HTML
49 lines
1.7 KiB
HTML
<html lang="en">
|
|
|
|
<head>
|
|
<meta charset="utf-8">
|
|
<meta name="viewport" content="width=device-width" />
|
|
<title>Next-gen Kaldi WebAssembly with sherpa-onnx for Speaker Diarization</title>
|
|
<style>
|
|
h1,div {
|
|
text-align: center;
|
|
}
|
|
textarea {
|
|
width:100%;
|
|
}
|
|
</style>
|
|
</head>
|
|
|
|
<body>
|
|
<h1>
|
|
Next-gen Kaldi + WebAssembly<br/>
|
|
Speaker Diarization <br> with <a href="https://github.com/k2-fsa/sherpa-onnx">sherpa-onnx</a>
|
|
</h1>
|
|
<div>
|
|
<span id="hint">Loading model ... ...</span>
|
|
<br/>
|
|
<br/>
|
|
<label for="avatar">Choose a wav file:</label>
|
|
<input type="file" id="file" accept=".wav" onchange="onFileChange()" disabled></input>
|
|
<br/>
|
|
<br/>
|
|
<label for="numClusters" id="numClustersID">Number of speakers: </label>
|
|
<input type="text" id="numClustersInputID" name="numClusters" value="-1" />
|
|
<br/>
|
|
<br/>
|
|
<label for="clusteringThreshold" id="thresholdID">Clustering threshold: </label>
|
|
<input type="text" id="thresholdInputID" name="clusteringThreshold" value="0.5" />
|
|
<br/>
|
|
<br/>
|
|
|
|
<textarea id="text" rows="10" placeholder="If you know the actual number of speakers in the input wave file, please provide it via Number of speakers. Otherwise, please leave Number of speakers to -1 and provide Clustering threshold instead. A larger threshold leads to fewer clusters, i.e., fewer speakers; a smaller threshold leads to more clusters, i.e., more speakers."></textarea>
|
|
<br/>
|
|
<br/>
|
|
<button id="startBtn" disabled>Start</button>
|
|
</div>
|
|
|
|
<script src="app-speaker-diarization.js"></script>
|
|
<script src="sherpa-onnx-speaker-diarization.js"></script>
|
|
<script src="sherpa-onnx-wasm-main-speaker-diarization.js"></script>
|
|
</body>
|