How to Transcribe a Panel Discussion (2026)

How to transcribe a panel discussion starts with a roster, not with a guess about Speaker 1. A panel is a moderator plus the people invited to talk: usually four to eight voices, on a stage or in a Zoom. Lapel mics into a mix, or one cloud recording. It is not a two-person interview, and it is not a focus group seated around a table. Those jobs have their own steps. Interview: how to transcribe an interview recording. Focus group: how to transcribe focus group discussions.

The panel difference is the moderator. They introduce, they hand the floor off, they cut people off when time runs out. Those cues are how you name the panelists. Do that before you trust the first speaker label the model prints.

What you are holding

You want a transcript a recap writer can quote without emailing you "who was this?" Named panelists. A moderator marked as the moderator. A time on the line you will publish.

You do not need every laugh. You do need the handoff ("Sam, you disagree") attached to Sam, and any sentence that will ship in the writeup or on the captions.

Stage and Zoom use the same later steps. The file you upload is what changes.

Step 1: Write the roster before you open the file

On paper, or in a note next to the editor:

• Moderator, full name
• Each panelist, in the order they were introduced
• Anyone who might speak once (a host who only does the welcome, an audience mic)

The run of show will not match Speaker 1, Speaker 2, Speaker 3. The model numbers voices, not the printed program. A hot lapel on the moderator often becomes the first label even when a panelist talks more. Knowing the names first stops you from locking the wrong person onto every turn.

If the panelists are public, use real names. If this is a closed briefing, roles are enough (Moderator, City, Finance). Decide that before you rename, so you do not leak a name into an export.

Step 2: Get one recording

Transcribe the mix a person would listen to.

• Stage: the board recording. Lapel mics or headset mics summed to one file are much kinder than a phone in row twelve. If all you have is the room mic, still use it. Budget a harder speaker pass. The cleanup is how to separate speakers on a panel recording.
• Zoom: download the cloud MP4 (or the local recording if cloud was off). One file. Gallery view is for watching. The audio mix is for the transcript. Path notes: transcribe Zoom meetings.

Skip the pile of phone videos unless the board file is missing. Six angles means six drafts and no agreement on who spoke.

Keep the original filename with the event and date. Panels from the same conference blur together by Thursday.

Step 3: Upload with speaker labels on

Open File Transcribe and upload the mix. Guest try works for a first pass with no account. Turn speaker labels on before it runs. A panel without labels is a wall of text, and you will not reconstruct four to eight people from paragraphs alone.

A 60 to 90 minute panel is a normal length. Processing follows the length of the audio, not the headcount. You can start the next step as soon as the draft is on screen. Plans, if this is one panel in a series: a monthly minute balance, Free about 50 minutes, Pro about $19 for about 2,000 minutes, Plus about $49 for about 5,000 minutes. Confirm on pricing.

Step 4: Use the moderator's cues before you rename anyone

Play the open. Almost every panel includes a sentence like "I'm Jordan, I'll moderate. With me: Priya, Sam, and Lena."

That sentence is the map. Listen to who says it. That voice is the moderator, even if the label says Speaker 4.

Then listen to the first handoff. "Priya, you have been in this market for a decade." The next long turn should be Priya. If the label flips to someone else halfway through her answer, do not rename the whole label yet. The model split her, or it gave the moderator's hot mic its own number. The next guide is the repair: separate speakers on a panel recording.

Do not do this from memory if you were backstage. The intro is on the tape. Use it.

Step 5: Proof handoffs and the lines you will quote

A full verbatim pass on a noisy panel is how the afternoon disappears. Proof in this order:

1. Introductions and handoffs. If the moderator assigns the floor to the wrong name in your transcript, every quote under that name is suspect.
2. Proper nouns. Companies, bills, product names. Panels are dense with them, and the model will guess.
3. Any line that will be published. Click the segment, listen, then fix. A recap can survive a fuzzy "yeah." It cannot survive a misquote.

Leave crosstalk marked if you cannot hear it. Inventing the buried half of an overlap is worse than a gap. Audience questions can wait until the named panel is stable. If a new voice appears, label it Audience rather than forcing it onto a panelist.

Step 6: Export the writeup or the captions

• Article or internal recap: download TXT or DOCX with speaker labels on. The names you set are what the writer sees.
• Video of the panel: export SRT or VTT after you sign in on a plan that includes captions (Pro or Plus). Read the captions against the picture. A paragraph that is fine in a recap is too long on screen.
• Both: same transcript. Export twice. You do not re-upload.

If the file is an interview with one guest, you used the wrong guide. Two voices are interview recordings. If the room was a focus group and the speakers should stay anonymous, use the focus group steps instead of putting legal names on participants.

How this stays different from the other two

An interview is two people and a quote that has to hold up. You rename Speaker 1 and Speaker 2 and you are mostly done.

A focus group is several participants whose overlap is part of the data, often anonymized, with a moderator fishing for reaction. Crosstalk is the analysis.

A panel is performed. People wait, then they talk, then the moderator points at someone else. Your job is to follow the pointing. Name them from the intro. Then fix the places the mix lied to the model.

FAQ

How many speakers can a panel transcript handle?

Four to eight, plus a moderator, is the normal range. The draft will number them. You still have to match numbers to the roster. More overlap means more listening, not a different upload.

Should I transcribe each lapel mic as its own file?

Only if you never got a mix. A single board mix or Zoom recording is the file to upload. Separate tracks mean you are assembling the panel by hand afterward.

Is a Zoom panel different from a stage panel?

The upload is the same. Zoom is one cloud MP4. A stage is the board mix, ideally with lapels. Both still need the moderator's intro before you trust Speaker 1.

What if two panelists talk at once?

Leave the line, listen, and fix attribution in a second pass. That pass is how to separate speakers on a panel recording. Do not smooth it into one person to make the doc look tidy.

Name them, then export

Upload the mix on File Transcribe. Play the introduction. Rename only after the handoffs match the roster. Then export the recap, or the captions, from the names you actually checked.

Related: Separate speakers on a panel · Zoom meetings · Interview recording steps · Focus group steps · Pricing

More guides

• Transcription guides
• Try File Transcribe free
• Transcript format guide
• Best transcription software
• Transcribe Zoom meetings

Further reading

• Zoom cloud recording help
• W3C captions overview