Correcting speakers
How to assign, reassign, and manage speakers, and how Avocet's matching improves over time.
Avocet diarizes the other side of the call into unlabeled voices like "SPEAKER_2." You tell it who that is once, and it remembers.
The meeting view
Open a finished meeting to see its transcript, grouped by speaker and time. Each speaker uses the same tint slot in the transcript, speaker map, and editor. Unassigned voices keep their raw label and a dashed chip. Blank names show ? in their initials chip.
Drag the top edge of the player bar up or down to give the speaker lanes more or less room. The edge lights up when you hover over it. You can also Tab to the resize handle and press Up or Down to change its height; Home makes it smallest and End makes it largest. Double-click the edge to return to automatic sizing. Avocet remembers your height across meetings and restarts. If you shrink the window, the bar fits the available space and returns to your chosen height when you enlarge the window.
The player bar's speed chip changes playback from 0.5x to 3.0x while preserving pitch with native macOS playback. The same default is used in later meetings and can be changed in Settings > Playback.
Playback is unavailable while a recording starts, runs, pauses, or stops. If a meeting is playing when recording starts, it pauses. The Play and Listen controls are disabled with the reason “Playback is off while Avocet is recording.” You can still seek through the meeting, but seeking does not start audio. Playback does not resume when recording ends.

An unassigned voice with a strong enough match shows an inline suggestion with a percentage. Click Accept if it is correct, or Other speaker to pick someone else. Next unassigned jumps you straight to the next voice that still needs a name.
Assigning and reassigning
Select words in the transcript and click the Reassign icon in the menu that appears, click a speaker's name chip, or press R to open the speaker editor for the sentence you last clicked. Microphone words are always yours, so Reassign stays disabled when you select words on your microphone track. Dictionary remains available for those words. A selection that spans your words and the other side's shows no actions; select within one side instead.
The menu has five icons: Reassign, Join, Correct, Dictionary and Delete. Hover an icon to see its name; a greyed icon says why it is unavailable. To join two sentences from the same confirmed speaker into one, select across the end of the first and the start of the next, then click Join. If one of your own sentences sits between them, Join still works, and the other icons stay greyed with the reason.
The speaker editor keeps Tab and Shift+Tab inside it. Esc clears your search first, then closes the editor on the next press. Closing returns focus to the speaker control that opened it. While a change saves, Esc and the backdrop leave the editor open.

The editor ranks your most likely known speakers first, with a match score for each. The selected speaker has a check mark, separate from the accent highlight on the current row. Choose one, or type a name to create a new speaker. You can apply the change to just that segment, the whole paragraph, or every matching segment in the meeting. If only part of a segment is wrong, drag to select just those words and reassign the selection.
Unassign removes a name from a segment and puts it back to its raw label.
Retrying transcription
Retry runs the same steps on the same audio, so it gives the same speakers unless you enter a speaker count. In the retry confirmation, “Speakers other than you (optional)” counts remote people on the system track. FluidAudio uses it as a maximum, so the result can contain fewer speakers. Leave it blank for automatic detection.
Retry also processes your microphone again. Text you corrected, split or deleted, and text you restored, stays. Other microphone text is replaced by the new result.
When your microphone was nearly silent for a stretch of your transcript, Avocet hides the text there because it is usually a recognition mistake. The words stay in your database and the audio is kept. This checks volume only, so loud background noise can still produce text. Meetings you recorded before this change are not rechecked until you retry them.
When your Mac's speakers play the call out loud, your microphone can hear the other side, and the same words land on your lines. Avocet hides the words on your microphone that repeat what the other side said at that moment, and the audio is kept. Retry checks them again.
The Speakers page
Every person Avocet has recognized shows up on the Speakers page, with how many meetings they appear in and how recently. Open a speaker to rename them, add aliases, merge two entries that turned out to be the same person, split a wrongly merged one apart, or delete a speaker who has no segments left.
How matching learns
Every time you confirm a name, Avocet keeps that voice sample as trusted evidence for that speaker. Recognizing someone with enough confidence requires at least two meetings and 20 trusted voice samples, so accuracy builds up over your first few calls with them.
Once a speaker has enough evidence, Avocet can assign them automatically, without waiting for you to accept a suggestion. If you correct an automatic assignment, Avocet remembers that near-miss and avoids repeating it for that voice, while still learning the assignment you chose.
Renaming a speaker changes their name everywhere it appears and regenerates every affected export automatically.