As of 13 August 2026, AI can only partly edit a podcast episode for a client.

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, power-user skill, and roughly 2 hours until you can act on the result.

   How to actually do it

   1. Open your audio editor and create a new project, keeping the original recordings in a separate untouched folder.
   2. Gather the cleanest recordings, the transcript, the client’s written brief, the required delivery format, intro and outro files, music licences and any list of words or sections that must remain.
   3. Paste the brief and transcript into an AI tool with the audio files if supported, using the prompt above, and ask it to separate definite edits from suggestions.
   4. Open each timestamp in the audio editor and apply only the definite cuts, checking that speech is not clipped and that edits do not change the speaker’s meaning.
   5. Listen to the complete episode with headphones and speakers, compare every change with the transcript and brief, and check dialogue levels, background noise, music transitions and the beginning and ending.
   6. Export the requested file format, open the exported file in a new player, listen through it again and compare its duration and contents with the client’s requirements.
   7. Send the finished file and a concise change summary to the client, retaining the project file and original recordings for revisions.

   Prompt

   Act as an assistant to a freelance podcast editor. Use the attached audio files, transcript and client brief to prepare an edit decision list for one podcast episode.
   
   Client brief:
   [PASTE THE CLIENT BRIEF]
   
   Episode details:
   - Desired final duration: [INSERT OR SAY UNKNOWN]
   - Required delivery format: [INSERT OR SAY UNKNOWN]
   - Editing style: [INSERT OR SAY UNKNOWN]
   - Intro, outro, adverts or music instructions: [INSERT OR SAY NONE]
   - Words, names or topics that must not be removed: [LIST OR SAY NONE]
   
   Work only from the supplied files and brief. Do not invent timestamps, speaker names, words, music licences or technical facts. Identify exact timestamps for pauses, filler words, repeated phrases, false starts, obvious mistakes, unwanted sounds and sections the brief asks to remove. Preserve meaning and natural conversation unless the brief says otherwise. Separate definite edits from suggestions and flag anything that needs my decision.
   
   Return:
   1. A chronological edit decision list with timestamps, quoted words where available, the proposed action and a short reason.
   2. A list of continuity, pronunciation and fact checks for me to make.
   3. A mixing and export checklist suited to the requested delivery format, without inventing loudness or file requirements.
   4. A short client-facing summary of the proposed changes.
   
   If your environment can directly edit and export the supplied audio, apply only the definite edits, keep the original files untouched, and report every applied change. Otherwise, produce the decision list and checklist only.

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