Mobius generates a transcript from any audio file dropped into the matter file space. The transcription runs automatically and produces a text file you can use as the input for Notes of Interview and other AI workflows.
## Supported file types
Mobius transcribes audio-only files, including MP3, M4A, and WAV. **Video files are not supported.** If you have a video recording (for example, an MP4 from Microsoft Teams), export the audio track first and upload that. The transcription model is purpose-built for spoken interview audio and consistently outperforms Teams' built-in transcription in terms of accuracy.
## Uploading an audio file
Open the matter, go to the file space, and drag the audio file in (or use the upload button in the top right corner). Transcription begins automatically — there is no separate workflow to trigger. The file appears in the file list with a status indicator while it processes.
## Processing time
As a rough guide, a 2.5-hour interview takes around 10–12 minutes to transcribe. Longer files take proportionally longer. The transcription runs on our server, so you can close the tab and come back to it without worrying about losing your work.
## Improving accuracy with keywords
Names, places, and case-specific terms are the most common sources of transcription error. Add these at the matter level (or, if relevant across cases, at the personal or organisation level), and they will be applied to every transcription run on the matter. Adding an interviewee’s name to the keyword list before uploading the interview is the single most effective step you can take to reduce manual correction later.