Podcast to text, straight from a link

No audio file to find, download or upload. Paste a link or search for the show, pick the episode, and Podskrift fetches the audio and transcribes it.

How it works

  1. Paste a link or search

    Spotify, Apple Podcasts, an RSS feed or a direct audio URL all work.

  2. Pick the episode

    You see the episode list with lengths before anything is charged.

  3. Get the text

    Plain text for reading and quoting, .srt with timestamps for subtitles.

Questions

Do I need to upload the audio file?

No. Podskrift downloads the episode from the podcast’s public feed. If you only have a file, host it somewhere and paste the direct audio URL.

How accurate is it?

It uses modern AI speech recognition, which is good on clear speech. Crosstalk, heavy accents and music beds reduce accuracy, and there is no speaker labelling yet.

How long does it take?

Usually a few minutes for an hour of audio. You can close the page; the job keeps going.

Can I use the transcript in ChatGPT or Claude?

Yes. Connect Podskrift as an MCP connector and ask for summaries or quotes from episodes, or paste the text yourself.

Which languages work best?

28 languages in the picker. Name the spoken language instead of auto-detect for short or accented audio. Norwegian has its own page under “More ways to transcribe”.

What does it cost?

Start free, no card needed. After that, minutes are a one-time purchase: from $1.50 for 60 minutes, cheaper per minute the more you buy (1,200 min for $11), or just the minutes one episode needs. No subscription. You can also add your own OpenAI API key and pay OpenAI directly instead; that is optional.

What do I get back?

The full transcript as plain text (.txt) and as SubRip subtitles (.srt) with timestamps, plus a reading view in the browser. Transcripts are private to your account unless you create a share link.

Which languages work?

28 languages, transcribed with modern speech recognition. Auto-detect is the default; choosing the language yourself helps on short or accented audio.