The Multilingual Code Switch enables the processing of audio content that contains multiple languages by continuously detecting the spoken language and switching between the corresponding language models.

​
Configuration

This feature is implicitly enabled by not specifying the language parameter. To disable this feature, set the language parameter to a specific language (e.g. language=en).

​
Results

Each word in the transcription-response will have a language field that indicates the detected language of the word.

{
  "transcription": " Good afternoon. Bună seara.",
  "words": [
    {
      "word": " Good",
      "start": 1199,
      "end": 1520,
      "confidence": 1.0,
      "language": "en",
      "metadata": {}
    },
    {
      "word": " afternoon.",
      "start": 1520,
      "end": 2020,
      "confidence": 0.99853516,
      "language": "en",
      "metadata": {}
    },
    {
      "word": " Bună",
      "start": 2159,
      "end": 2320,
      "confidence": 1.0,
      "language": "ro",
      "metadata": {}
    },
    {
      "word": " seara.",
      "start": 2320,
      "end": 2480,
      "confidence": 1.0,
      "language": "ro",
      "metadata": {}
    }
  ],
  "start": 0,
  "end": 3071,
  "metadata": {},
  "utterance": false
}
Get API accessUtterance End