Transcriptions

Convert audio to text using speech-to-text models.

Endpoint

POST /v1/audio/transcriptions

Authentication

Authorization: Bearer gw_prod_...

Request

Headers

Header	Required	Description
Authorization	Yes	Bearer token with API key
Content-Type	Yes	multipart/form-data

Body Parameters

Parameter	Type	Required	Description
file	file	Yes	Audio file (mp3, mp4, mpeg, mpga, m4a, wav, webm)
model	string	Yes	Model ID: whisper-1, voxtral-mini-latest
language	string	No	ISO-639-1 language code (e.g., en, es, fr)
prompt	string	No	Optional text to guide transcription style
response_format	string	No	json, text, srt, verbose_json, vtt
temperature	number	No	Sampling temperature (0-1)
timestamp_granularities	array	No	word, segment (verbose_json only)

Examples

Basic Transcription

curl -X POST https://api.gateflow.ai/v1/audio/transcriptions \
  -H "Authorization: Bearer gw_prod_..." \
  -F "file=@audio.mp3" \
  -F "model=whisper-1"

Python

import openai

client = openai.OpenAI(
    base_url="https://api.gateflow.ai/v1",
    api_key="gw_prod_..."
)

with open("audio.mp3", "rb") as audio_file:
    transcript = client.audio.transcriptions.create(
        model="whisper-1",
        file=audio_file
    )

print(transcript.text)

With Language Hint

transcript = client.audio.transcriptions.create(
    model="whisper-1",
    file=audio_file,
    language="es"  # Spanish
)

With Timestamps

transcript = client.audio.transcriptions.create(
    model="whisper-1",
    file=audio_file,
    response_format="verbose_json",
    timestamp_granularities=["word", "segment"]
)

for segment in transcript.segments:
    print(f"[{segment.start:.2f}s - {segment.end:.2f}s] {segment.text}")

Streaming Transcription

import httpx

async with httpx.AsyncClient() as client:
    with open("audio.mp3", "rb") as f:
        async with client.stream(
            "POST",
            "https://api.gateflow.ai/v1/audio/transcriptions",
            headers={"Authorization": "Bearer gw_prod_..."},
            files={"file": f},
            data={"model": "voxtral-mini-latest", "stream": "true"}
        ) as response:
            async for chunk in response.aiter_text():
                print(chunk)

Response

JSON Format (Default)

{
  "text": "Hello, how can I help you today?"
}

Verbose JSON Format

{
  "task": "transcribe",
  "language": "english",
  "duration": 5.5,
  "text": "Hello, how can I help you today?",
  "segments": [
    {
      "id": 0,
      "seek": 0,
      "start": 0.0,
      "end": 2.5,
      "text": "Hello, how can I help you today?",
      "tokens": [50364, 2425, 11, 577, 393, 286, 854, 291, 965, 30],
      "temperature": 0.0,
      "avg_logprob": -0.25,
      "compression_ratio": 1.2,
      "no_speech_prob": 0.01
    }
  ],
  "words": [
    {"word": "Hello", "start": 0.0, "end": 0.5},
    {"word": "how", "start": 0.6, "end": 0.8},
    {"word": "can", "start": 0.9, "end": 1.1},
    {"word": "I", "start": 1.2, "end": 1.3},
    {"word": "help", "start": 1.4, "end": 1.6},
    {"word": "you", "start": 1.7, "end": 1.9},
    {"word": "today", "start": 2.0, "end": 2.5}
  ]
}

SRT Format

1
00:00:00,000 --> 00:00:02,500
Hello, how can I help you today?

VTT Format

WEBVTT

00:00:00.000 --> 00:00:02.500
Hello, how can I help you today?

GateFlow Extensions

Provider Selection

transcript = client.audio.transcriptions.create(
    model="auto",  # Let GateFlow choose
    file=audio_file,
    extra_body={
        "gateflow": {
            "prefer": "quality"  # or "speed" or "cost"
        }
    }
)

Fallbacks

transcript = client.audio.transcriptions.create(
    model="whisper-1",
    file=audio_file,
    extra_body={
        "gateflow": {
            "fallbacks": ["voxtral-mini-latest"]
        }
    }
)

Supported Models

Model	Provider	Languages	Max File Size	Best For
whisper-1	OpenAI	50+	25MB	General transcription, translation
voxtral-mini-latest	Mistral	100+	25MB	Multilingual, diarization
voxtral-mini-2602	Mistral	100+	25MB	Multilingual, diarization
gemini-2.5-flash	Google	100+	25MB	Fast multilingual
gemini-2.5-pro	Google	100+	25MB	High accuracy

Mistral Voxtral Features

Mistral Voxtral models support additional features:

Parameter	Type	Description
diarize	boolean	Enable speaker identification
context_bias	string	Comma-separated domain-specific words (max 100)

# Transcription with speaker diarization
transcript = client.audio.transcriptions.create(
    model="voxtral-mini-latest",
    file=audio_file,
    extra_body={
        "diarize": True,
        "context_bias": "GateFlow,API,authentication"
    }
)

Voxtral Limitations

Mistral Voxtral does not support translation. Use OpenAI Whisper for audio translation to English.

Errors

Code	Description
400	Invalid file format or parameters
401	Invalid API key
413	File too large (max 25MB)
429	Rate limit exceeded
500	Provider error

See Also

• Speech Synthesis - Text to speech
• Audio Providers - Provider configuration
• Streaming Speech - Real-time audio