Create Realtime Transcription Session

Create a WebSocket session for real-time audio transcription.

POST /v1/realtime/transcription_sessions

cURL

curl https://gateway.bud.studio/v1/realtime/transcription_sessions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "whisper-1",
    "language": "en"
  }'

200

{
  "id": "session_xyz789",
  "object": "realtime.transcription_session",
  "created_at": 1699000000,
  "model": "whisper-1",
  "language": "en",
  "ws_url": "wss://gateway.bud.studio/v1/realtime/transcription_sessions/session_xyz789",
  "expires_at": 1699003600
}

cURL

curl https://gateway.bud.studio/v1/realtime/transcription_sessions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "whisper-1",
    "language": "en"
  }'

200

{
  "id": "session_xyz789",
  "object": "realtime.transcription_session",
  "created_at": 1699000000,
  "model": "whisper-1",
  "language": "en",
  "ws_url": "wss://gateway.bud.studio/v1/realtime/transcription_sessions/session_xyz789",
  "expires_at": 1699003600
}

Headers

Parameter	Type	Required	Description
Authorization	string	Yes	Bearer authentication header

Body

Parameter	Type	Required	Description
model	string	Yes	Transcription model identifier
language	string	No	ISO-639-1 language code for transcription
response_format	string	No	Format: json, text. Default: json
temperature	float	No	Sampling temperature (0.0 to 1.0)

Supported Providers

OpenAI

Real-time audio streaming with GPT-4o

Azure OpenAI

Enterprise real-time API with low latency