In one sentence
Audio recordings in the DMS become a transcript with speakers and timestamps when you click, and the case file records what is said in them as a person's statement, not as an established fact.
How it works
A client sends three WhatsApp voice messages in which she describes a conversation with her landlord. You upload the files to the case file, right-click to choose “Transcribe (speakers + timestamps)” and confirm the notice. The transcript is created as a separate document in the case file, with lines such as “[02:14] Speaker 1: …”. Under “Name the speakers” you assign the voices to parties in the case file. Common formats such as MP3, WAV, M4A and WhatsApp voice messages are recognized, as well as videos such as MP4 or MOV; for large videos, the browser extracts only the audio track. Entire WhatsApp chat exports can be imported, including their voice messages.
Statement instead of fact
Facts from a transcript record who said them. They appear under “Statements from conversations”, and a click plays the passage in the recording. Contradictions are also checked with regard to the speaker: if the same person says different things in two places, this is flagged separately.
Costs and data protection
Transcription runs only when you start it; nothing is transcribed automatically. It is billed per started minute from your AI quota, and only if the transcript is created. You see the costs beforehand in the dialog. Audio cannot be pseudonymized: the recording goes unchanged to our transcription service provider Mistral. That is why the dialog points this out before the first transfer, and you decide for each recording.
What it does not do
- Speaker separation distinguishes voices but does not identify persons. Names and numbers can be misheard; check important passages against the audio.
- A recording may be up to 50 MB in size and up to three hours long.
- Extracting the audio track of large videos requires a current browser such as Chrome, Edge or Safari.
- Whether a recording was made lawfully and may be used is for you to assess.
More: Document management, in the manual Transcribing audio recordings and Statements from conversations.
