Home · Business · Admin & Office · Minutes & transcription

PARTLY

As of 13 August 2026, AI can only partly add speaker names to a meeting transcript.

This still needs a person who signs their name to it.

Can you do it?

5 minutesto a draft.

30 minutesto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ita colleague

What the alternative costsThe supplied tool information gives no price for a manual or specialist transcription alternative.

If this goes wrong: the transcript attributes a decision or comment to the wrong person and the error is copied into minutes, actions or an internal record.

What to actually do

  1. Hand it to a person

    The route this page recommends

    A person who owns the outcome does this end to end, worth it when the failure is dear.

  2. Use a tool built for this

    Second choice

  3. Do it yourself

    The distant third

    A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

    How to actually do it

    1. Open the transcript and, if available, the meeting recording in separate windows.
    2. Gather the final attendee list, roles, any known absences, and reliable notes about who chaired or presented each agenda item.
    3. Paste the participant list, meeting context, transcript and any audio notes into a chatbot using the supplied prompt.
    4. Ask the chatbot to produce the labelled transcript, confidence table and unresolved sections without changing the spoken wording.
    5. Play the recording at each flagged speaker change and compare the voice, timestamp and surrounding exchange with the proposed name.
    6. Send the corrected transcript to a meeting participant who was present, asking them to confirm disputed names before using it for minutes, actions or decisions.

    Prompt

    Add speaker names to the meeting transcript below. Use only the participant information and transcript evidence provided. Do not guess a person's identity from writing style, job title or the order of the participant list. If you cannot identify a speaker confidently, use a neutral label such as Speaker 1 or Unknown speaker and explain why. Preserve the transcript wording, timestamps and order. Mark overlapping or unclear speech rather than silently rewriting it. Return: 1) the labelled transcript, 2) a table of each speaker label, proposed name, confidence level and evidence, and 3) a list of every section that needs checking against the audio or meeting participants. Participant list: [paste names and roles here]. Voice or attendance notes: [paste any reliable notes here]. Meeting context: [paste agenda or relevant context here]. Transcript: [paste transcript here]. Audio or recording notes, if available: [paste them here].

    Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

What it gets wrong

  • AI cannot reliably identify a person from text alone when several participants have similar speaking styles or roles.
  • It cannot recover a name that is absent from the transcript, attendee list or recording context.
  • It cannot settle overlapping or indistinct speech without a human checking the recording.
  • It cannot know whether an attribution is sensitive or disputed in your workplace.
  • It cannot take responsibility for an incorrect record sent to colleagues.

What caps this at PARTLY: context depth, verification cost and consent and privacy.

How we scored this

Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.

AxisScore (0–2)
Output2
Inputs1
Verification1
Liability1
Effort delta2
Total7 / 10

The methodology and its thresholds are published in full.

FAQ

Can ChatGPT identify who said what in a transcript?
It can propose speaker names when you provide the transcript, participant list and useful context. It cannot reliably identify people from text alone, so use neutral labels for uncertain sections and check them against the recording.
How accurate is AI at adding speaker names to a transcript?
Accuracy depends on the quality of the transcript, the number of speakers, the audio and the information you provide. Similar voices, interruptions and missing introductions can produce confident-looking but incorrect names.
Can AI label speakers without the audio?
It can make tentative assignments from names, roles, turn-taking and meeting context, but those assignments are not proof of identity. Without audio or a participant who can confirm the result, leave uncertain passages as Speaker 1 or Unknown speaker.
Is it safe to upload a work transcript to AI?
Check your employer's policy before uploading it, especially if it contains personal, confidential or commercially sensitive information. Remove unnecessary identifying details and use an approved meeting or transcription tool where your organisation requires one.

Nearby answers

All Minutes & transcription rulings

Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.

The newsletter

AI news, new answers and product picks, straight to your inbox.