Speaker Diarizationbeta

Watch the demo

Watch it on YouTube.

Speaker Diarization

Speaker diarization separates a conversation into different voices and gives each voice a label. This makes a multi-person transcript easier to read and helps you connect decisions, questions, and action items to the right person.

Think of it like placing name tags on voices - Meetily first groups lines that sound like the same person, then you listen and replace labels such as Speaker 1 with names you recognize.

Speaker diarization is currently in Beta. It is ready to use, and we are actively improving how consistently voices are separated across different meeting conditions. You can review and adjust speaker labels whenever needed.

Enable it at Settings > PRO > Speaker Diarization. It is off by default. When it is off, Meetily uses Host and Guest labels where available; there is no additional setting for that fallback.

At a glance

Quick Start

1. Open Settings > PRO > Speaker Diarization and turn the feature on.
2. Record a meeting, import audio, or start retranscription.
3. When processing finishes, Meetily opens the Meeting page for a live recording. For an imported or retranscribed meeting, select it in the sidebar.
4. Open Transcript, then select Speakers, use Listen to recognize each person, and enter their names.
5. Select Done to save your speaker names and any staged changes.

Enable → Process meeting → Identify voices → Review labels → Done

Choose How to Use It

Live recording

Record your meeting as usual. Meetily identifies voices after you stop the recording, so final speaker labels do not appear during the live session. When processing finishes, Meetily opens the Meeting page. Open Transcript and select Speakers.

Audio import

Turn speaker diarization on before selecting Import Audio. Meetily can identify voices while you continue using the app. When it finishes, select the imported meeting in the sidebar, open Transcript, and select Speakers.

Retranscription

Turn speaker diarization on before selecting Retranscribe. Meetily creates a new speaker result and restores existing names from that meeting only when it can reliably recognize the same voice. When it finishes, select the meeting in the sidebar, open Transcript, and select Speakers.

Review and Name People

1. For a live recording, Meetily opens the Meeting page after processing. For an imported or retranscribed meeting, select it in the sidebar.
2. Open Transcript, then select Speakers in the bottom toolbar.
3. Select Listen beside a detected voice to hear a short sample.
4. Select the current label, then enter the person's name.
5. Repeat for the people you recognize and select Done.

Names must be different from one another, contain at least one character, and be no more than 100 characters long.

What gets saved, and when

Improve Speaker Results

Set the expected number of people

Meetily normally chooses the number of voices automatically. If the result has too many or too few people, you can give it a number and try again.

1. Open Speakers and find Participants.
2. Choose Auto, or select a number from 1 to 10.
3. Select Re-run.
4. Continue with another meeting or another task while Meetily identifies the voices again.
5. Return to the transcript and review the updated list when it finishes.

Meetily can still return fewer voices than the number you selected when the recording does not contain enough clear, distinct speech. Short replies, overlapping speech, background noise, echo, distant microphones, and similar voices can all affect the result.

Keep Meetily running

You can move to another meeting, inspect other details, or close the main window to the system tray while identification continues. You do not need to keep the progress panel open.

Merge Duplicate People

Sometimes one person's voice appears as two or more entries. Open the Merge tab to combine those entries into one person.

1. Select two or more entries that belong to the same person.
2. Use Keep as to choose the name that should remain.
3. Select Merge to stage the change.
4. Review the staged result and select Undo if it is not correct.
5. Select Done when you are ready to save.

If only one speaker remains, there is nothing else to combine, so merging is unavailable.

Correct One Transcript Line

Use a line-level correction when the people are identified correctly overall but one transcript segment has the wrong name.

1. Find the line with the incorrect speaker.
2. Select the speaker-label dropdown beside that line.
3. Choose the correct person from the available speakers.

The change saves immediately. You do not need to open the speaker panel or select Done.

Use the Finished Transcript

If speaker identification cannot find useful labels, your saved transcript and recording remain available. You can still read, copy, export, or summarize the meeting using the labels Meetily has available.

How It Works

Meetily labels voices without changing what anyone said. It separates audible voices, attaches a temporary label to each group, and lets you confirm the people before using the result elsewhere.

Enable feature → Process audio → Separate voices → Review people → Save labels → Use transcript

• Enable feature: Turn speaker diarization on before starting the recording, import, or retranscription.
• Process audio: Meetily transcribes the meeting and listens for changes between voices.
• Separate voices: Lines that sound like the same person are grouped under a speaker label.
• Review people: You can listen, add names, set the expected count, merge duplicates, or correct one line.
• Use transcript: Saved names appear in the transcript, copied or exported text, and newly generated summaries.

Real-World Examples

Example 1: Reviewing a live team meeting

"I want the project discussion to show who made each decision."

Setup: Enable diarization > Record meeting > Stop > Listen and name people > Done. The completed transcript shows the names you saved beside their lines.

Example 2: Labeling an imported interview

"I imported an interview and need the interviewer and guest clearly labeled."

Setup: Enable diarization > Import Audio > Wait for processing > Open meeting > Review speakers. Rename the two voices before copying or exporting the transcript.

Example 3: Combining duplicate entries

"The same person appears twice in this transcript."

Setup: Open Speakers > Merge > Select duplicates > Choose Keep as > Merge > Done. The person's lines appear under the single name you kept.