Interview transcription with speaker labels

An interview transcript is only useful if you can tell who said what. Skeleton Fingers labels speakers on the fast lane, where our servers process an hour of audio in seconds.

Who said what

Interviews are conversations, and a wall of unattributed text is slow to work with. On the fast lane, Skeleton Fingers labels each speaker, so the interviewer's questions sit apart from the answers and you can find a quote without replaying the recording. Language is detected automatically, so an interview in another language does not need any setup.

Choosing a lane for interviews

If you promised a source that the recording would stay private, or the subject matter is sensitive, the browser lane keeps the audio on your own machine. You will add speaker names yourself, but the recording is never uploaded. For everything else, the fast lane saves the most time.

Getting a clean interview recording

Transcription quality starts with the recording. A few habits make a real difference:

• Keep the microphone close to both voices. A phone on the table between two people works better than one across the room.
• Pick a quiet room. Cafes, air conditioners and traffic compete with the voices you care about.
• Avoid talking over each other. Overlapping speech is hard for any transcriber, human or software, and it muddles speaker labels.
• For remote interviews, record the call itself. Zoom, Teams and Google Meet can all record, and that file is usually cleaner than a phone held up to a speaker.
• Say names and spellings on the recording. Asking a guest to spell their name at the start saves a lookup later.

What it costs

A free account transcribes interviews up to 30 minutes free, 3 a day. Longer interviews use pack hours for their full length: an hour pack is $5 for 3 fast hours, paid once, and the hours never expire. That pack covers four 45 minute interviews. If you interview every week, Pro is $9.99 a month for 10 fast hours, and pack hours stack on top. The browser lane is always free.

Before you publish a quote

Automatic transcripts are a fast first draft, not a final record. Before you quote someone, play back the moment and check the wording, and double check names, places and numbers. Speaker labels can also swap when two voices sound alike, so confirm who said a line before you attribute it.