Languages

In multilingual environments such as customer support centers handling calls in multiple languages, media companies processing international content, or platforms accepting user-generated audio, you can't always know the language of an audio file before processing it. Sending audio with the wrong language code produces poor or failed transcripts. SubQ supports explicit language selection and automatic detection to handle both known and unknown language scenarios.

Supported languages

SubQ supports both English and Spanish transcription:

Language codes follow the BCP-47 format.

Setting the language

When you know the language of your audio, pass the language query parameter (or its alias lang) in the transcription request:

curl -X POST "https://stt-api.subq.ai/v1/listen?language=es" \
  -H "Authorization: Bearer YOUR_SUBQ_API_KEY" \
  --data-binary @audio.wav

# WebSocket URL with language parameter
wss://stt-api.subq.ai/v1/listen?language=es&interim_results=true

When no language parameter is provided, the API defaults to English.

Automatic language detection

When you don't know the language of your audio files, enable automatic detection. SubQ analyzes the audio's acoustic features and phonetic patterns to identify the spoken language, then routes the audio through the corresponding transcription model. For SubQ APIs to identify the spoken language automatically, set detect_language=true as follows:

curl -X POST "https://stt-api.subq.ai/v1/listen?detect_language=true" \
  -H "Authorization: Bearer YOUR_SUBQ_API_KEY" \
  --data-binary @audio.wav

You can also pass language=auto as a shorthand:

curl -X POST "https://stt-api.subq.ai/v1/listen?language=auto" \
  -H "Authorization: Bearer YOUR_SUBQ_API_KEY" \
  --data-binary @audio.wav

When language detection is enabled, the response includes a detected_language field so your application can act on the result. For example, routing a support ticket to an agent who speaks the detected language.

{
  "results": {
    "channels": [{
      "detected_language": "es",
      "alternatives": [{
        "transcript": "Hola, ¿cómo estás?",
        "confidence": 0.91
      }]
    }]
  }
}

Next steps

• Pre-recorded quickstart
• Streaming quickstart