Hanzo_dev
7695dd0be6
Pipeline completo: URL → Whisper → GPT-4o → pgvector → Supabase Frontend Vue 3 + Tailwind, Backend Express + Vercel serverless functions
37 lines
1.3 KiB
JavaScript
37 lines
1.3 KiB
JavaScript
// ============================================================
|
|
// TRANSCRIPTOR — OpenAI Whisper
|
|
// Descarga el audio desde la URL y lo transcribe
|
|
// ============================================================
|
|
import OpenAI from 'openai'
|
|
|
|
const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY })
|
|
|
|
/**
|
|
* @param {string} audioUrl URL directa del MP3 (de Social Download API)
|
|
* @param {string} idioma Código de idioma: 'es', 'en', 'pt', etc.
|
|
* @returns {string} Transcripción completa del audio
|
|
*/
|
|
export async function transcribir(audioUrl, idioma = 'es') {
|
|
// Descargar el audio desde la URL del CDN
|
|
const audioResponse = await fetch(audioUrl)
|
|
if (!audioResponse.ok) {
|
|
throw new Error(`Error al descargar audio: ${audioResponse.status}`)
|
|
}
|
|
|
|
const audioBuffer = await audioResponse.arrayBuffer()
|
|
const audioFile = new File([audioBuffer], 'audio.mp3', { type: 'audio/mpeg' })
|
|
|
|
const transcripcion = await openai.audio.transcriptions.create({
|
|
file: audioFile,
|
|
model: 'whisper-1',
|
|
language: idioma === 'otro' ? undefined : idioma, // auto-detect si es 'otro'
|
|
response_format: 'text',
|
|
})
|
|
|
|
if (!transcripcion || transcripcion.trim().length < 10) {
|
|
throw new Error('Whisper no pudo transcribir el audio (resultado vacío o muy corto)')
|
|
}
|
|
|
|
return transcripcion.trim()
|
|
}
|