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.
To transcribe an interview, upload the recording to Kai The Scribe, tick Label different speakers, and click Start Transcription. The transcript comes back as one paragraph per turn, labelled Speaker 1, Speaker 2 and so on. Swap those labels for real names, then check every quote you plan to use against the audio.
Interviews are the recordings people most want in writing. A journalist needs exact quotes. A researcher needs to code what each participant said. A hiring manager wants to compare candidates side by side. A podcaster wants a guest episode in a form readers can skim. All four jobs start with the same thing: a transcript that shows who said what. This guide covers how to record an interview so it transcribes well, how to run it through the tool, and how to clean up the result into something you can quote from.
Most transcript problems are recording problems. A few habits during the interview save more cleanup time than anything you do afterwards.
If you need to record people, follow the consent rules that apply where you and they are, and your organization's own policy. Those rules vary by place, so check with the relevant authority or your legal team rather than relying on a general guide.
There are five steps. The labels below match the tool.
With Label different speakers ticked, the tool groups the words into turns and starts each one with a label. A two person interview reads like this:
Speaker 1: Thanks for making the time. Can you start with how the project began?
Speaker 2: Sure. It started as a side project in our team, and it grew from there.The heading above the text says with speaker labels when speaker data came back. The text is punctuated and formatted, but it carries no timestamps. If you leave the box unticked, you get the same words as one block of text with no turns.
A raw transcript is a draft. Three quick passes turn it into a usable record.
For an article, pull the strongest three or four quotes into a separate document first, then write around them. For research, paste the named transcript into your analysis tool so each code stays tied to the person who said it. For a podcast guest episode, the podcast transcription guide covers turning the text into show notes. For a hiring panel, keep each candidate's transcript in one folder so the panel reads the same answers.
Recording quality matters more than any setting. The guide to recording audio for transcription goes deeper on microphones, rooms and file formats.
Yes, when you tick Label different speakers. The transcript comes back as paragraphs that start with Speaker 1:, Speaker 2: and so on, one paragraph per turn. The tool does not know names, so replace each label with the person’s name using find and replace.
Yes, as long as you have the recording as a file. Export the audio from your call app or recorder, then upload it. A share link to a playback page will not work, because the tool needs the media file itself.
No. The output is plain text with speaker paragraphs, not a timestamped or caption file. To find a quote in the audio, search the transcript for a nearby phrase and scrub to roughly the same point in your player.
Use it to find the quote, then confirm the exact words against the recording before you publish. Names, numbers, and crosstalk are where machine transcripts slip most often.
Every account gets 60 free minutes a month, which covers one hour-long interview. Pro is $19 a month for unlimited transcripts. Minutes count the length of the recording.