Skip to main content
Audio translation converts speech from any language to English text: 
from pathlib import Path

response = client.audio.translations.create(
    file=Path("french_audio.mp3"),
    model="openai/whisper-large-v3",
)
print(f"English translation: {response.text}")

import { createReadStream } from 'fs';

const response = await together.audio.translations.create({
  file: createReadStream('french_audio.mp3'),
  model: 'openai/whisper-large-v3',
});
console.log(`English translation: ${response.text}`);

​
Translation with context

from pathlib import Path

response = client.audio.translations.create(
    file=Path("business_meeting_spanish.mp3"),
    model="openai/whisper-large-v3",
    prompt="This is a business meeting discussing quarterly sales results.",
)

​
Limits and errors

/v1/audio/translations shares the same code path as transcription: the 500 MB direct-upload cap, 1 GB URL-fetch cap, and 4-hour duration cap all apply, and the same error codes are returned. For audio above 500 MB, submit an HTTPS URL on the file field instead. See Limits and Errors and troubleshooting.