Setting the Acoustic or Language Model when transcribing

    You can use a different Acoustic Model and/or Language Model which better matches your data in order to get higher accuracy.

    Choosing an Acoustic Model

    Below is a table of all the publicly available Acoustic Models. You can learn more about Acoustic Models and how it impacts accuracy on the Choosing a Model guide.

    Model Use Case Parameter Name
    Default      Best for all types of data and accents. English only. "assemblyai_default"
    Australian English Tailored for Australian accented English "assemblyai_en_au"
    South African English Tailored for South African accented English "assemblyai_en_za"
    Custom Tailored for your data. Contact us for more info: support@assemblyai.com

    To swap the Acoustic Model, set the "acoustic_model" parameter to the name of the Acoustic Model you want to use when submitting transcription jobs to the API. For example:

    curl --request POST \
      --url https://api.assemblyai.com/v2/transcript \
      --header 'authorization: YOUR-API-TOKEN' \
      --header 'content-type: application/json' \
      --data '
        {
            "audio_url": "https://s3-us-west-2.amazonaws.com/blog.assemblyai.com/audio/8-7-2018-post/7510.mp3",
            "acoustic_model": "assemblyai_en_au"
        }'

    Choosing a Language Model

    Below is a table of all the publicly available Language Models. You can learn more about Language Models and how it impacts accuracy on the Choosing a Model guide.

    Model Use Case Parameter Name
    Default Best for conversational data like phone calls and meetings. "assemblyai_default"
    Media Best for broadcast news and radio and at recognizing proper nouns, this model has a much larger vocabulary than the default model. "assemblyai_media"

    To specify the Language Model when submitting transcription jobs, set the "language_model" parameter to the name of the Language Model you want to use. For example:

    curl --request POST \
      --url https://api.assemblyai.com/v2/transcript \
      --header 'authorization: YOUR-API-TOKEN' \
      --header 'content-type: application/json' \
      --data '
        {
            "audio_url": "https://s3-us-west-2.amazonaws.com/blog.assemblyai.com/audio/8-7-2018-post/7510.mp3",
            "language_model": "assemblyai_media"
        }'