kunkunsh-kunkun-ext-speech-to-text/deno-src/transcribe.ts

import OpenAI from 'openai';

const apiKey = Deno.env.get('OPENAI_API_KEY');
const openai = new OpenAI({ apiKey });

// expect a audio.m4a file in the current directory
const fileData = await Deno.readFile('./chinese.m4a');

// Convert to a File (filename is required)
const file = new File([fileData], 'audio.m4a', { type: 'audio/m4a' });

const transcription = await openai.audio.transcriptions.create({
	file: file,
	model: 'whisper-1',
	language: 'zh' // this is optional but helps the model
});
// const transcription = await openai.audio.transcriptions.create({
// 	file: file,
// 	model: 'whisper-1',
// 	language: 'en' // this is optional but helps the model
// });

console.log(transcription);