# Speakers are mixed up or mislabeled

> Why speaker separation merges or splits voices, how to fix labels with Rename speaker, and how notetaker meetings get real names.

Updated: 2026-08-16 · Canonical: https://usecanto.com/help/troubleshooting/speakers-mislabeled/

Speaker separation works from voice characteristics alone, so similar-sounding voices, speakerphone audio, and people talking over each other get merged or swapped. Wrong names are a one-tap fix with **Rename speaker**; meetings recorded by the notetaker use real participant names instead of guesses.

## Why do speakers get merged or swapped?

Canto tells speakers apart by how their voices sound, not by who they are. That breaks down predictably:

- **Similar voices** — two people with close pitch and cadence can be merged into one speaker.
- **Speakerphone and conference-room audio** — everyone's voice passes through the same distant microphone and room echo, which flattens the differences separation relies on. This is the most common cause of merged speakers.
- **Overlapping speech** — when people talk over each other, the overlap is attributed to whichever voice dominates.
- **One person, two labels** — a speaker who moves around the room, or whose audio quality shifts mid-meeting, can be split into two.

More on how separation works: [How does speaker separation work?](https://usecanto.com/help/transcription/speaker-separation/)

## How do I fix a wrong name?

1. Open the recording and tap a speaker label in the transcript.
2. Choose **Rename speaker** and type the person's name.
3. The name applies to every line from that speaker, for everyone with access.

Full steps: [Rename speakers](https://usecanto.com/help/transcription/rename-speakers/). Renaming fixes labels, not attribution — if two people were merged into one voice, renaming can't split their lines apart. For lines that were transcribed under the wrong speaker, you can still [edit the transcript text](https://usecanto.com/help/transcription/edit-a-transcript/) itself.

## Why are notetaker meetings labeled better?

When the notetaker records a meeting, the meeting platform reports who is speaking when, so Canto labels turns with real participant names — no voice guessing involved. Names can still look off in two cases: someone dialed in by phone or joined under a shared account, and a name the platform reports (like a bare tenant account name) may differ from what you'd call the person. **Rename speaker** fixes those the same way.

If accurate names matter for a Teams, Meet, or Zoom meeting, send the notetaker rather than recording the call through your device's microphone — a device recording of a video call is exactly the speakerphone case above. See [What is the notetaker?](https://usecanto.com/help/notetaker/what-is-the-notetaker/)

## How do I prevent it next time?

- Put the phone in the middle of the table, closer to the quieter speakers — see [Get better audio quality](https://usecanto.com/help/recording/microphone-tips/).
- Avoid recording a speakerphone from across the room; one clear microphone near the conversation beats a distant one.
- Encourage one-at-a-time speaking in meetings where attribution matters.

Related articles:

- [Rename speakers in a transcript](https://usecanto.com/help/transcription/rename-speakers/)
- [How does Canto tell speakers apart?](https://usecanto.com/help/transcription/speaker-separation/)
- [How do I get the best audio quality?](https://usecanto.com/help/recording/microphone-tips/)