Free podcast and interview transcription

Short video transcripts work best line by line. Long talk does not: a conversation split into a hundred short lines is hard to read and hard to quote. This tool groups the speech model's segments into paragraphs, starting a new one at a long pause or after a sentence once a paragraph has run for a while.

Each paragraph keeps the timestamp where it starts, so you can jump back to that point in your editor to cut a clip or check a quote.

Interviews are harder for speech models than a single voice. A few things help:

• Use a segment where people take turns rather than talk over each other. Crosstalk is where most mistakes happen.
• If you recorded each person on their own track, transcribe the mixed track for reading, or each track on its own to know exactly who said what.
• Cut intro music and ad reads first. Music under speech lowers accuracy.
• Check guest names, company names and jargon. Those are the words a model most often gets wrong.

Find the two or three moments that stand on their own without the rest of the episode. Those are your clip candidates. Pull a quote for the show notes, a newsletter or a carousel. Then take the strongest opening line you find and test it in the hook analyzer before you cut the clip.