Transcription (STT)

Turn speech into text (speech-to-text, STT) with Whisper, GPT-4o Transcribe, Deepgram, Voxtral and other transcription models. The endpoint follows OpenAI's POST /audio/transcriptions, so client.audio.transcriptions.create(...) works unchanged.

POST https://api14.com/api/v1/audio/transcriptions

Request

Send multipart/form-data with the audio file (what the OpenAI SDKs do), or a JSON body.

The request body limit is 25 MB by default. Larger files return 413 payload_too_large, split long recordings into chunks.

Multipart with cURL

curl https://api14.com/api/v1/audio/transcriptions \
  -H "Authorization: Bearer sk-api14-..." \
  -F model=openai/whisper-1 \
  -F file=@meeting.mp3 \
  -F language=en

JSON body

JSON is accepted too, with the audio as base64 for models that support it:

{
  "model": "openai/whisper-1",
  "input_audio": { "data": "<base64>", "format": "mp3" },
  "language": "en"
}

Example

import fs from "node:fs";
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api14.com/api/v1",
  apiKey: process.env.API14_API_KEY,
});

const result = await client.audio.transcriptions.create({
  model: "openai/whisper-1",
  file: fs.createReadStream("audio.mp3"),
});
console.log(result.text);

Response

{
  "text": "Thanks everyone for joining. Let's start with the roadmap.",
  "usage": { "type": "duration", "seconds": 42, "cost": 0.0042 }
}

With response_format: "text", srt or vtt the body is plain text instead of JSON.

Pricing

Transcription is billed at the model's listed price, shown as "$ / minute" (or per token for token-priced models). The charge is returned in usage.cost and debited from your USD balance.

Errors

See Errors.