Home · Personal · Writing & Creative · Music & audio
As of 13 August 2026, AI can transcribe your podcast recording.
This still needs a person who signs their name to it.
Can you do it?
5 minutesto a draft.
1 hourto something you’d act on.
Cost, all in£0
Skill needednone
Who has to check ityou
What the alternative costsDescript is a purpose-built alternative that edits audio and video by editing the AI transcript.
If this goes wrong: the transcript contains wrong words or names, and you need to correct it against the recording before publishing.
What to actually do
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.
Do it yourself
Second choiceA chat interface, no skill needed, and roughly 1 hour until you can act on the result.
How to actually do it
- Open a transcription-capable AI tool such as ChatGPT or Descript and start a new transcription project or chat.
- Gather the original podcast recording and any notes containing the episode title, speaker names, guest names, places, organisations and specialist terms.
- Upload the recording and paste the prompt, replacing the bracketed context slot with your notes.
- Ask the tool to produce the complete transcript with speaker labels, timestamps and [inaudible] markers rather than a summary.
- Open the transcript beside the recording and listen through it, checking every speaker name, quotation, number, place, technical term and marked unclear passage.
- Correct the transcript in your document, then search for any remaining AI-added wording or summary text and remove it.
- Save the checked transcript with the episode title and send or publish it only after comparing the final text with the audio.
Prompt
Transcribe the attached podcast recording exactly as spoken. Use separate speaker labels and add timestamps at natural intervals and whenever the speaker changes. Preserve hesitations and meaningful repeated words, but do not add information, tidy up wording or guess at unclear speech. Mark anything you cannot hear confidently as [inaudible 00:00:00] using the nearest timestamp. Use these speaker names if they are clear from the recording: [SPEAKER 1], [SPEAKER 2], [GUEST]. Check names, places, organisations and technical terms against this context before finalising: [PASTE CONTEXT HERE]. Return the transcript in plain text with a short list of any passages, names or terms that need me to check against the audio.
Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.
Use a tool built for this
The distant third
What it gets wrong
- AI cannot reliably identify every speaker when voices overlap, sound similar or are recorded unevenly.
- It cannot know whether an unfamiliar name, place or technical term has been transcribed correctly without context from you.
- It cannot decide whether a hesitation, false start or repeated phrase matters for the version you want to publish.
- It cannot verify the transcript without you listening back to the recording, and a full check takes time.
- It cannot take responsibility for errors in a transcript published under your name.
Even on a YES, the friction has a name: verification cost and context depth.
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.
| Axis | Score (0–2) |
|---|---|
| Output | 2 |
| Inputs | 2 |
| Verification | 1 |
| Liability | 2 |
| Effort delta | 2 |
| Total | 9 / 10 |
The methodology and its thresholds are published in full.
FAQ
- Can ChatGPT transcribe an audio file?
- Yes, if the version you are using accepts the recording upload, it can produce a transcript from spoken audio. Give it speaker names and context, and tell it to mark unclear words instead of guessing.
- How accurate is AI transcription for podcasts?
- It is most dependable when the recording is clear, speakers do not talk over one another and names or specialist terms are supplied. You still need to listen back because a plausible-looking transcript can contain wrong words, names and numbers.
- Can AI tell who is speaking in a podcast?
- It can often separate speakers when their voices are distinct and the conversation is clear. It can mislabel speakers during interruptions or overlapping speech, so compare each label with the recording.
- Do I need to check an AI-generated podcast transcript?
- Yes. Check names, quotations, numbers, places, technical terms and every passage marked unclear against the audio before publishing or sending it to someone else.
Nearby answers
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.