Documentación API

Upload an audio or video file for transcription with speaker diarization, language detection, and word-level timestamps.

Parámetros de solicitud

Enviar como multipart/form-data

Respuesta (JSON)

{
  "text": "Hello, welcome to the meeting...",
  "language": "en",
  "duration": 125.4,
  "segments": [
    {
      "start": 0.0,
      "end": 3.2,
      "text": "Hello, welcome to the meeting.",
      "speaker": "Speaker 1",
      "confidence": 0.95,
      "words": [
        {"word": "Hello", "start": 0.0, "end": 0.4},
        {"word": "welcome", "start": 0.5, "end": 0.9}
      ]
    }
  ],
  "speakers": ["Speaker 1", "Speaker 2"]
}

Summarize transcript text using an on-device LLM. No data leaves our servers.

Cuerpo de la solicitud (JSON)

Respuesta

{
  "summary": "The team discussed Q3 revenue growth of 15%...",
  "style": "brief",
  "model": "qwen2.5-1.5b-instruct"
}

Real-time speech-to-text via WebSocket. Send raw PCM audio (16-bit, 16kHz, mono) and receive transcription updates instantly.

Protocolo

1. Connect to wss://api.stt.ai/v1/stream
2. Send JSON config: {"language": "en", "model": "large-v3-turbo"}
3. Wait for {"status": "ready"}
4. Stream raw PCM Int16 audio chunks (binary frames)
5. Receive JSON updates: {"text": "...", "partial": "..."}
6. Send {"action": "stop"} to finalize

Mensajes del servidor

List all available transcription models. No authentication required.

{
  "models": [
    {"id": "large-v3-turbo", "name": "Whisper Large V3 Turbo", "languages": 99, "speed": "fast"},
    {"id": "large-v3", "name": "Whisper Large V3", "languages": 99, "speed": "standard"},
    {"id": "medium", "name": "Whisper Medium", "languages": 99, "speed": "fast"},
    {"id": "small", "name": "Whisper Small", "languages": 99, "speed": "very_fast"}
  ]
}

List all supported languages with ISO codes. No authentication required.

{
  "languages": [
    {"code": "en", "name": "english"},
    {"code": "es", "name": "spanish"},
    {"code": "fr", "name": "french"},
    ...
  ]
}

Check GPU and API health. No authentication required.

{
  "status": "ok",
  "gpu_available": true,
  "gpu_name": "NVIDIA A100",
  "gpu_memory_mb": 8188
}

cURL

# Transcribe a file
curl -X POST https://api.stt.ai/v1/transcribe \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "file=@meeting.mp3" \
  -F "model=large-v3-turbo" \
  -F "language=auto" \
  -F "diarize=true"

# Get SRT subtitles
curl -X POST https://api.stt.ai/v1/transcribe \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "file=@video.mp4" \
  -F "response_format=srt" \
  -o subtitles.srt

# Summarize text
curl -X POST https://api.stt.ai/v1/summarize \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text": "Full transcript text here...", "style": "brief"}'

# List models (no auth needed)
curl https://api.stt.ai/v1/models

# Health check
curl https://api.stt.ai/health

Python

import requests

API_KEY = "YOUR_API_KEY"
BASE = "https://api.stt.ai"

# Transcribe a file
with open("meeting.mp3", "rb") as f:
    resp = requests.post(
        f"{BASE}/v1/transcribe",
        headers={"Authorization": f"Bearer {API_KEY}"},
        files={"file": ("meeting.mp3", f, "audio/mpeg")},
        data={"model": "large-v3-turbo", "language": "auto", "diarize": "true"},
    )

result = resp.json()
print(f"Language: {result['language']}, Duration: {result['duration']:.1f}s")

for seg in result["segments"]:
    print(f"[{seg['start']:.1f}s] {seg.get('speaker', '')}: {seg['text']}")

# Summarize the transcript
summary = requests.post(
    f"{BASE}/v1/summarize",
    headers={"Authorization": f"Bearer {API_KEY}"},
    json={"text": result["text"], "style": "bullet_points"},
).json()
print(summary["summary"])

Node.js

const fs = require("fs");
const FormData = require("form-data");

const API_KEY = "YOUR_API_KEY";
const BASE = "https://api.stt.ai";

async function transcribe(filePath) {
  const form = new FormData();
  form.append("file", fs.createReadStream(filePath));
  form.append("model", "large-v3-turbo");
  form.append("language", "auto");
  form.append("diarize", "true");

  const resp = await fetch(`${BASE}/v1/transcribe`, {
    method: "POST",
    headers: { Authorization: `Bearer ${API_KEY}`, ...form.getHeaders() },
    body: form,
  });

  const result = await resp.json();
  console.log(`Duration: ${result.duration.toFixed(1)}s`);

  for (const seg of result.segments) {
    console.log(`[${seg.start.toFixed(1)}s] ${seg.speaker}: ${seg.text}`);
  }
  return result;
}

transcribe("meeting.mp3");

WebSocket (Browser)

// Real-time transcription from microphone
const ws = new WebSocket("wss://api.stt.ai/v1/stream");
ws.binaryType = "arraybuffer";

ws.onopen = () => {
  ws.send(JSON.stringify({ language: "auto", model: "large-v3-turbo" }));
};

ws.onmessage = (event) => {
  const data = JSON.parse(event.data);
  if (data.status === "ready") {
    startStreaming(); // Begin sending audio
  }
  if (data.text) console.log("Final:", data.text);
  if (data.partial) console.log("Partial:", data.partial);
};

async function startStreaming() {
  const stream = await navigator.mediaDevices.getUserMedia({ audio: true });
  const ctx = new AudioContext({ sampleRate: 16000 });
  const source = ctx.createMediaStreamSource(stream);
  const processor = ctx.createScriptProcessor(4096, 1, 1);

  processor.onaudioprocess = (e) => {
    const float32 = e.inputBuffer.getChannelData(0);
    const int16 = new Int16Array(float32.length);
    for (let i = 0; i < float32.length; i++) {
      int16[i] = Math.max(-32768, Math.min(32767, float32[i] * 32767));
    }
    if (ws.readyState === 1) ws.send(int16.buffer);
  };

  source.connect(processor);
  processor.connect(ctx.destination);
}

// Stop: ws.send(JSON.stringify({ action: "stop" }));