By the end of this guide you will have turned a podcast episode into a text transcript with speaker labels, checked against the audio and exported in a format you can publish. The job is done by an AI speech-to-text tool: you upload the audio file (or record inside the tool), it writes the transcript, and you proofread it.
If you want to transcribe a podcast with AI, the real decision is where the work happens: inside a podcast editor or recording studio, in a dedicated transcription app, on your own computer, or through an API you call yourself. This guide walks through each option and compares six tools from our Audio and voice category and neighbouring ones.
Key takeaways
- Upload the finished episode audio to a transcription tool, choose the spoken language, switch on speaker labels, then proofread the draft against the recording.
- The main decision is whether you want transcription as part of editing and recording (Descript, Riverside), as a standalone app, offline on your own PC, or as an API for developers.
- Free plans exist but are small: for example, 60 media minutes a month on Descript, 300 transcription minutes a month on Otter.ai, or one 30-minute transcription a day on EKHOS AI.
- AI transcripts are drafts. Names, jargon and quotes need a human check, especially for legal, medical or academic content.
What you need before you start
- The episode audio. Use the final mix or the raw recording as an MP3 or WAV file. Exporting from your editor is usually enough.
- An account or an install. Check each tool's sign-up terms. The offline option, EKHOS AI, needs a Windows PC and a one-time sign-in. The developer option needs some comfort with code.
- A list of names. Guest names, brand names and technical terms help you proofread quickly.
- Time. Setup and a first pass take under an hour for a typical episode. Careful proofreading takes longer for long or technical shows.
- Difficulty. Easy for the web apps, moderate for the API.
- Consent. If a guest or caller has not agreed to being recorded and transcribed, sort that out before you publish their words.
Which kind of AI tool fits this job
Pick by the situation you are in rather than by feature lists.
A podcast editor, where the transcript is also the edit. Descript transcribes your recording and lets you change the media by editing the text, so deleting a word removes that moment from the timeline. Its listing names transcription in 25 languages, filler-word removal and captions and show notes. If you want to cut the episode and transcribe it in one place, this is the shape of tool to look at.
A recording studio that transcribes as you record. Riverside records each participant locally in separate tracks, then offers AI transcription in 100+ languages and a text-based editor. It also lists show notes and clips as outputs. It fits shows that record remote guests and want the transcript to arrive from the same service.
A transcription app for finished files. Good Tape transcribes audio and video in 100+ languages, puts speaker labels on every transcript, and has an editor with audio playback and notes. Otter.ai is an AI meeting assistant that also imports audio and video files, identifies speakers and exports to mp3, txt, pdf, docx and srt. Its transcription covers English, Spanish, French, German, Japanese and Chinese, so check your show's language first.
An offline app when the audio must not leave your machine. EKHOS AI is a Windows app that transcribes on your own computer, so recordings and transcripts are not sent to a server. It supports 98 languages and common formats such as MP3, WAV and MP4.
An API for developers and pipelines. Deepgram offers speech-to-text through an API for recorded and live audio, with speaker diarization and automatic language detection. It is aimed at developers who will wire transcription into their own publishing workflow.
Step 1: Export a clean audio file
Good input gives you a better draft, so start from the cleanest recording you have.
- Finish your edit and export the episode as MP3 or WAV.
- Keep every voice on one file if you can. If you recorded separate tracks, mix them down first, unless your tool takes tracks directly.
- Remove long stretches of music or ads that you do not need transcribed.
- Note the file length and size, because some free plans cap minutes or imports.
Check it worked by playing the first and last minute of the exported file.
Step 2: Choose a tool to transcribe your podcast with AI
You end this step with one tool picked and an account or install ready.
- Decide where the audio may go: a hosted service, your own PC only, or your own code.
- Check the free plan against your episode length. Good Tape's Free plan allows 3 tapes a month up to 30 minutes each, and EKHOS AI's Free plan allows one 30-minute transcription a day. Otter.ai's Basic plan has 300 monthly transcription minutes but only 3 lifetime file imports, which matters if you upload finished files.
- Check the language list covers your show.
- Use the comparison table further down to see pricing and stated limits side by side.
Check it worked by confirming your episode length fits within the plan you chose.
Step 3: Upload the episode and set language and speakers
The goal is a first draft with the right language and separate speakers.
- Sign in, or open the app, and start a new transcription or project.
- Upload your file. If you record in Riverside, the recording is already in the tool. On Premium, EKHOS AI can also record and transcribe what your speakers play.
- Select the spoken language, or use automatic detection if the tool offers it. Deepgram lists automatic language detection.
- Turn on speaker labels or diarization. Good Tape labels speakers on every transcript. On EKHOS AI, speaker identification is a Premium feature. Otter.ai lists speaker identification, and Deepgram lists speaker diarization.
Check it worked by seeing the file appear in your project list with a processing status.
Step 4: Run the transcription and review the speaker labels
Let the tool finish, then read through the draft for structural problems before you fix wording.
- Wait for the transcript to complete.
- Open the transcript editor and skim the whole episode once.
- Check that each speaker label matches the right person, and rename generic labels to real names. Good Tape lets you rename a speaker once and update the whole transcript.
- Look for places where two people talk over each other, since these are where labels slip.
Check it worked by confirming that a few spot-checked lines are attributed to the right person.
Step 5: Proofread against the audio
Accuracy is your job here. The AI gives you a draft, not a finished record.
- Play the audio alongside the text. Good Tape's editor has audio playback, and EKHOS AI's tracks editor plays the media in sync with the text. In Descript and Riverside you edit the transcript in a text-based editor.
- Fix guest names, brand names, acronyms and numbers first, using your list.
- Correct punctuation and paragraph breaks so the text reads naturally.
- Re-listen to every direct quote you plan to publish.
- Have a person check anything legal, medical, financial or academic before it goes out.
Check it worked by reading one section aloud against the recording without finding new errors.
Step 6: Export and publish the transcript
Save the transcript in the format your site or workflow expects, and keep a copy.
- Export to plain text for a web page or show notes. Otter.ai lists exports to mp3, txt, pdf, docx and srt, and EKHOS AI exports to Word, PDF and text, with Word and PDF on Premium.
- Keep your own copy of the transcript and the audio. Credits, minutes and imports on free plans can run out, and some do not roll over.
- Add the transcript to the episode page, and credit the speakers.
Check it worked by opening the exported file in the app you will publish it from.
Step 7: Reuse the transcript
Once the text exists, it can feed other work.
- Ask for show notes or a summary. Descript lists captions and show notes, Riverside lists show notes and posts, and Good Tape and Otter.ai produce AI summaries with action points.
- For short clips, Descript and Riverside both list clip creation from long recordings.
- Search your back catalogue. Good Tape has an Explore feature for topics and questions across an archive, and Otter.ai has an AI chat that answers questions across many meetings.
Check it worked by publishing one clip or summary and comparing it with the transcript.
Tools for this job compared
| Tool | Good fit for | Pricing | Stated limits |
|---|---|---|---|
| Descript | Editing a podcast by editing its transcript (web) | Freemium, paid from $16/mo (checked 24 Oct 2026) | Free plan: 60 media minutes a month and 100 one-time AI credits; unused minutes and credits do not roll over |
| Otter.ai | Transcribing imported files and calls, with summaries (web, iOS, Android, API) | Freemium, paid from $8.33/mo (checked 24 Oct 2026) | Basic: 300 monthly transcription minutes and 3 lifetime file imports; minutes do not roll over |
| Riverside | Recording remote guests in separate tracks, then editing the transcript (web) | Freemium, paid from $24/mo (checked 24 Oct 2026) | Free plan: 2 hours of multi-track recording, up to 720p and a watermark; AI Credits run out |
| Good Tape | Hosted transcripts with speaker labels and an editor (web, iOS, Android, API) | Freemium, paid from €16/mo (checked 24 Oct 2026) | Free plan: 3 tapes per month up to 30 minutes each; Pro includes 20 hours a month; some features marked as coming soon |
| EKHOS AI | Private, offline transcription on Windows | Freemium, paid from $9/mo (checked 24 Oct 2026) | Windows only; Free plan allows one 30-minute transcription a day; speaker identification and Word and PDF export are on Premium |
| Deepgram | Developers transcribing recorded or live audio by API | Freemium | Developer-oriented, integrated through APIs and SDKs; concurrency limits vary by plan |
Prices and plan limits change, so confirm the current terms on each vendor's site before you subscribe.
Common mistakes to avoid
- Trusting the first draft. Speech recognition stumbles on names, jargon and crosstalk. Fix it by proofreading against the audio before publishing.
- Picking a plan that cannot fit the episode. Free tiers are capped by length, by count or by monthly minutes, so a long episode can hit a limit halfway through. Fix it by comparing your episode length with the plan limits before you upload.
- Skipping speaker labels. A wall of unlabelled text is hard to read and easy to misquote. Fix it by turning on speaker labels in the tool, and check whether that is a paid feature, as it is on EKHOS AI.
- Uploading material you may not share. Unreleased episodes and private interviews may carry confidentiality or consent obligations. Fix it by getting guest consent and, where needed, choosing an offline tool such as EKHOS AI.
- Ignoring language coverage. Tools list different languages, from six on Otter.ai to 100+ on Good Tape and Riverside. Fix it by checking your show's language before you commit to a plan.
How we put this together
This is an evidence-based guide built from each tool's listing in our directory, which is compiled from the vendor's own site and pricing page. It is not a review based on using the tools. We don't rank tools, and nobody paid to be included. Some outbound links may be affiliate links; see our disclosure.
Frequently asked questions
Is there a free way to transcribe a podcast with AI?
Yes, within limits. Descript's Free plan includes 60 media minutes a month, Otter.ai's Basic plan has 300 monthly transcription minutes, Good Tape's Free plan allows 3 tapes a month up to 30 minutes each, and EKHOS AI's Free plan allows one 30-minute transcription a day. Deepgram's Pay As You Go plan starts with $200 in free credit. A full-length episode may need a paid plan.
How long does it take to transcribe a podcast?
The listings do not publish a standard processing time, so run a short test file first. Plan on the proofreading taking most of your effort: you need to listen through the audio and correct names, terms and speaker labels.
Can AI transcribe a podcast accurately?
It produces a useful draft, not a guaranteed accurate record. Accuracy depends on audio quality, accents, crosstalk and specialist vocabulary. Always proofread against the audio, and have a person review any transcript used for legal, medical, financial or academic purposes.
Which tool is good for transcribing on a phone?
Good Tape lists iOS and Android apps, including a recorder app, and Otter.ai lists iOS and Android apps where you can follow transcripts live. Descript, Riverside and Deepgram list web or API access, and EKHOS AI is Windows only. Check each vendor's site for the current app features before you rely on them.
Can I keep my audio private?
EKHOS AI runs entirely on your computer, so recordings and transcripts are never sent to a server. Good Tape describes EU servers, encryption and no use of files to train models. Read each vendor's data terms before uploading confidential audio.
Compare more options in our Audio and voice category.
