Find out who said what
With the Pro plan, ask for the Actor positions format in a sitting’s Summary tab: it answers the question “who said what?”.
How attribution is done, and why a blank is better than a mistake.
In detail
How attribution works
The transcription first separates the voices, without naming them. A second pass then matches each voice to a name, drawing on the list of the chamber’s parliamentarians and government members, on the agenda and on introductory formulas of the “la parole est à madame” kind. The name retained must match a name on that list exactly; it then links to the parliamentarian’s profile from the transcript block.
The rule that governs the result
The system prefers to attribute nothing rather than get it wrong. If two people could match, the field stays empty. Interruptions, reported remarks and vote results do not count as interventions.
It is a deliberate choice: in our line of work, a sure attribution is worth far more than an approximate one. When the field is left open, the timecode takes you back to the picture to identify the speaker.