Audio & dictation

Dictating

Click Dictate (in Search & chat) to speak a question instead of typing it — see Search & chat for the button states.

Who said what

Recorded conversations show up as transcripts, grouped by speaker. Until you name someone, Selv shows a plain placeholder — "Unknown speaker (microphone)" for your own mic, or "Unknown speaker" plus the app or call it came from — rather than guessing a name.

Naming or fixing a speaker

Tap a speaker's name to open Identify this speaker: pick from voices Selv already knows, or type a new name. You choose how far the rename reaches — just that one line, every turn from that voice in this conversation, or that voice across every transcript.

If Selv attributed a line to the wrong person, or a badge shows "Uncertain speaker — tap to confirm," fixing it is the same rename action. If one block actually contains more than one person talking over each other, "This turn is multiple speakers — split it…" lets you mark where one speaker ends and the next begins. There's also a "Mark as noise (not speech)" toggle for lines that aren't really speech at all — it only changes how a line is labeled, never who it's attributed to.

Finding this

Open Recall → Recall tools → Recordings and speaker tools to browse transcripts and make these fixes.