When more than one person is on the other side of a call, talat tells their voices apart so you can follow who said what. This runs on every recording and there’s no switch to turn it off. Once you’ve named someone, talat can also recognise that person in future meetings; how eagerly it does that is your choice, in Settings → Speakers (covered below).
How voices appear in the transcript
The transcript reads as lines of speech, each with the speaker’s name shown at the start of the line:
- You are labelled You, with a pink accent.
- People talat already knows (from a previous meeting where you named them) can be recognised and shown with their name and a unique colour.
- People talat hasn’t heard before are labelled Unassigned, in a neutral colour. talat does not guess names for new voices; that’s your job.
The first meeting with someone new will have a lot of Unassigned lines. That’s expected. Once you name them, future meetings get easier.
Recognising people across meetings
Whether a known voice is recognised in later meetings is set under Settings → Speakers, with three choices:
- Off. talat never carries names between meetings; you name people fresh each time.
- Suggest. When a voice matches someone you’ve named before, talat surfaces them as a suggestion and you confirm before they’re added. This is the default.
- Automatic. A confident match is applied on its own, with no prompt.
The same tab has an option to assume the microphone is you at the start of a recording, either on calls or always, so your own lines are labelled from the first word.
Naming someone
Click the speaker name at the start of a line to open the picker. On a line talat hasn’t identified, that name reads Unassigned. Click it and either:
- Type a name and press Enter to create a new person.
- Pick someone you’ve named before from the list.
Clicking the speaker on a line always opens this picker, whether the line is still Unassigned or already has a name you want to change.
Names are global. A person you call Alice in one meeting will appear under Alice in every other meeting they’re in.
Helping talat recognise people next time
When you name or reassign a line, talat may store a short clip of that person’s voice as a voice print. Each person has up to three voice-print slots. In future meetings, talat compares unfamiliar voices against every stored voice print and matches them when one is close enough.
A few things you can do to make recognition more reliable:
- Name people early in the call. The longer talat has a name attached to a voice, the more material it has to learn from.
- Reassign mistakes as you spot them. Each correction can refresh or fill a voice-print slot, which sharpens future matching.
- Give it a few seconds of clean speech. A voice print needs at least seven seconds of audio to be stored. Short lines (“yeah”, “right”) won’t seed one.
Fixing incorrect attributions
Click the speaker name on any misattributed line and pick (or create) the correct person. The transcript updates immediately, every other line from that voice in the meeting moves with it, and talat may use the correction to improve a voice print for the target person. From the meeting’s side panel you can also reassign all of one person’s lines to another in a single step.
If talat keeps confusing two people, the cause is usually a single bad voice print. Open one of the affected people, listen back to their voice prints, and delete the worst-sounding one.
Managing voice prints
Open Speakers in the sidebar and pick a person to see their detail view. Their voice prints are listed under Voice prints. Each one shows:
- A play button, to listen back and confirm it really is them.
- The duration of the clip and when it was captured.
- A trash icon to delete the voice print.
Empty slots fill back up the next time talat captures eligible audio for that person, either from auto-matching or from a reassignment you make. Deleting a poor-quality voice print (crosstalk, background noise, the wrong person) is almost always better than keeping it; one bad voice print degrades matching more than a missing one.
The same detail view is where you add a person’s email addresses. Binding an email lets talat match that person to a calendar invitee’s address automatically when you connect a calendar.
The audio clip behind a voice print is only kept for a few days after a meeting. Once it ages out, the voice print still works for matching; you just can’t play it back to verify.
Merging two people into one
If the same person ends up split across two named people (a voice got matched to two different names, say), open the duplicate in Speakers, open the ⋯ menu at the top of their detail view, and choose Merge into…. Pick the correct person and confirm. Their lines move over and the duplicate is removed.
Deleting a person
On a person’s detail view, click the trash icon in the top right. talat asks where their lines should go:
- Reassign to another person. Pick someone from the list and every line moves to them.
- No one (unlink). Their lines stay in the transcript but go back to Unassigned.
Either way, the person’s voice prints are removed with them.
When recognition struggles
- Heavily overlapping speech. When two or more people talk over each other, even known voices can be misattributed.
- Bluetooth headsets in call mode. The other person’s mic drops to a low-fidelity codec, which makes them harder to recognise from one meeting to the next.
- Very short lines. One-word interjections don’t carry enough voice for a confident match.