This month we added a highly requested feature to the app.
The feature automatically transcribes audio recordings of conversations into written text. Transcription - writing out a recorded conversation such as an interview or confrontation meeting - is a difficult and time-consuming task. As a result, fully written-up transcripts of such conversations are rarely produced in practice.
Typically only a brief summary is written up, which can lead to disputes afterwards and still takes considerable time.
We have now added the option to have audio recordings automatically transcribed - a word-for-word transcript of the entire conversation, which can be shared with all parties immediately after processing, directly from the Montr app.
This is how it looks in the app, where the user can simply upload the audio recording:
Processing time is normally around 5 minutes per 20 minutes of audio, depending on the length of the recording. During busy periods it may take a little longer.
Once the transcript is ready, the user receives an email notification. An example of a completed transcript in the Montr app:
The transcript can be edited in the app before finalising - the output will never be 100% perfect and may contain occasional errors. Once finalised, a PDF is generated which can easily be shared with all participants.
The technology is based on the Whisper AI speech-to-text model. Privacy has been a key consideration in the implementation: the audio input is not stored in the model and is not used to further train it. Transcription takes place entirely within the EU. If you would like more information about this, please get in touch.
The model is still under active development, with performance continuing to improve and multilingual support also being added.
To use this feature, ask your application administrator to add the "audio" question type to a questionnaire - or consider creating a dedicated short questionnaire for it, as shown in the screenshot above.