Recognize different languages in one file

Answered

Comments

2 comments

  • Official comment
    Avatar
    Connor

    Hi Lars and Matt,

    Indeed an algorithm that is capable of transcribing multiple languages in the same audio file is beyond the current capabilities of the technology. The issue is we have completely different algorithms for each separate language and there's no system in place to recognise the language being spoken and then apply a specific algorithm only to that particular section of the audio file. 

    However, the alternative proposed by Matt may be possible with one of our newest features, Vocab Builder. Admittedly I haven't tried this myself, the feature is designed for technical jargon and custom words. You could theoretically fill this list up with words from other languages that frequently appear in your transcriptions.

    Hope this helps,


    Connor


  • Avatar
    Matt Yamashita

    I like this idea, but I imagine it would be difficult to implement. Another idea would be the ability to select a word, sentence, or paragraph in a transcript and designate it for transcription in a different language. Say you transcribe a mostly English interview, but there is occasional foreign language content. It would be nice if we could select just the foreign language content during review, mark it as whatever language it is, and have Trint automatically re-transcribe the selected content based on the language selected. 

Please sign in to leave a comment.