Transcription is live again: 60 free minutes a month, Pro is $19/mo. Other tools are coming back. Students: lecture notes still live at VocalScribe.
Transcribe a podcast episode by pasting the episode's direct MP3 link into Kai The Scribe, or uploading the file, and clicking Start Transcription. A 45 minute episode comes back as punctuated plain text in about a minute, ready to copy into show notes or download as .txt.
Podcast transcription is the job the tool was rebuilt for. Podcast transcription needs three things: a way to feed in a long audio file, clean punctuation so the text reads as sentences, and an output you can paste anywhere. The tool covers all three, and the first 60 minutes each month are free.
A transcript does four jobs for a show. Each job pays back the minute it takes to make.
There are five steps. The labels below match the tool exactly.
Every published episode has a direct file link, because that link is what podcast apps download. The direct file link lives in your RSS feed inside the enclosure tag of each item. Three ways to get it are below.
A player page from Spotify, Apple Podcasts, or YouTube is not a direct link. The tool fetches the URL you paste and expects audio bytes, so a player page returns an error. Paste the .mp3 link instead.
The output is one block of punctuated plain text. Sentences are capitalized, numbers are written as digits, and filler sounds are mostly dropped. A 45 minute conversational episode produces roughly 6,000 to 7,500 words. The summary card above the text shows the episode duration and the detected language, which is a quick check that the right file was processed.
Speakers come back numbered, not named. With Label different speakers ticked, each turn starts with Speaker 1:, Speaker 2: and so on. For a two person show, paste the text into your editor and use find and replace to swap the numbers for names.
Show notes come from the transcript in three passes. Pass one: search the text for the guest's name and the episode topic, and mark the five moments where the conversation turns. Pass two: write one sentence for each moment and add the timestamp from your audio editor. Pass three: paste the full transcript below the notes on the episode page, because the full text is what search engines index. The whole job takes 15 minutes for an hour of audio once the transcript exists.
A 45 minute episode pasted as a URL usually comes back in under a minute. An upload adds the time it takes your connection to send the file. The tool allows 60 seconds of processing per request, and episodes under an hour fit inside that window.
Yes, as numbered speakers. Tick Label different speakers and the transcript comes back as one paragraph per turn, starting Speaker 1:, Speaker 2: and so on. Swap the numbers for the host and guest names with find and replace.
Only if you have the episode file or its direct MP3 link. Spotify and Apple Podcasts pages are players, not files. Your own episodes have a direct link in your RSS feed. Other people's episodes need their permission and their file.
Plain text with punctuation, numbers, and sentence breaks. Copy it into your show notes editor or download it as a .txt file. It is not an SRT caption file, so it is not a drop-in for video captions.
Every account gets 60 free minutes a month, which covers one 45 minute episode or several short ones. Pro is $19 a month for weekly shows. Minutes count the length of the episode, not processing time.