whisper
Model ID: @cf/openai/whisper
Automatic speech recognition (ASR) system trained on 680,000 hours of multilingual and multitask supervised data
Properties
Task Type: Automatic Speech Recognition
Code Examples
Workers - TypeScript
import { Ai } from "@cloudflare/ai";
export interface Env { AI: Ai;
}
export default { async fetch(request: Request, env: Env) { const res: any = await fetch( "https://github.com/Azure-Samples/cognitive-services-speech-sdk/raw/master/samples/cpp/windows/console/samples/enrollment_audio_katie.wav" ); const blob = await res.arrayBuffer();
const ai = new Ai(env.AI); const input = { audio: [...new Uint8Array(blob)], };
const response = await ai.run( "@cf/openai/whisper", input );
return Response.json({ input: { audio: [] }, response }); },
};
curl
curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/run/@cf/openai/whisper \ -X POST \ -H "Authorization: Bearer $CLOUDFLARE_API_TOKEN" \ --data-binary @talking-llama.mp3 Response
Automatic speech recognition responses return both a single string text property with the audio transciption and an optional array of words with start and end timestamps if the model supports that.
Here’s an example of the output from the @cf/openai/whisper model:
{ "text": "It is a good day", "word_count": 5, "words": [ { "word": "It", "start": 0.5600000023841858, "end": 1 }, { "word": "is", "start": 1, "end": 1.100000023841858 }, { "word": "a", "start": 1.100000023841858, "end": 1.2200000286102295 }, { "word": "good", "start": 1.2200000286102295, "end": 1.3200000524520874 }, { "word": "day", "start": 1.3200000524520874, "end": 1.4600000381469727 } ]
}