Transcribe

Transcribe turns audio and video into text. It adds speaker labels and word timings. You can edit the transcript and export it as text, Word or subtitles.

Your transcripts

Transcribe opens on your transcripts, newest first, with each one's length, language, speakers and when it was made. Search them by file name or words in the transcript, and choose one to open it.

Transcribe a file

1. Open Transcribe in the sidebar.
2. Start a transcript in any of these ways:
   • drop an audio or video file anywhere on the list
   • choose Choose file in the bar at the top
   • choose New transcript (top right) or Record. In the window that opens, drop a file, choose Choose file, or choose Record to record with your microphone and Stop when you're done.
3. In the New transcript window, check the file name and length. The file plays below it. Choose Change to pick another.
4. Set Language, Key terms, Label speakers and Tag sounds if you need to.
5. Choose Transcribe. The estimate shows beside it.

The file uploads first (Uploading… N%). Choose Cancel to stop the upload; the window stays open until it's done or cancelled. Then the transcript opens and shows Transcribing…. You can leave the page. The transcript appears in the list when it's done.

Options

• Language: Detect automatically, or choose one of 31: English, Hindi, Telugu, Tamil, Kannada, Malayalam, Marathi, Bengali, Gujarati, Punjabi, Urdu, Arabic, Spanish, Portuguese, French, German, Italian, Dutch, Polish, Turkish, Russian, Ukrainian, Chinese, Japanese, Korean, Indonesian, Malay, Filipino, Vietnamese, Thai, Swahili.
• Label speakers: shows who said what. On by default.
• Tag sounds: adds tags like [laughs], [applause] and [music]. On by default.
• Key terms: names, brands and jargon, so they're spelled correctly. Press Enter (or type a comma) after each one. Choose × to remove one.

Work with the transcript

The transcript fits the window: the file's details, Find in transcript and the player at the top, the transcript below (it scrolls on its own), and a panel on the right with Export, View and Speakers tabs. All transcripts (top right) goes back to the list.

• Play from a word: click any word. The current word is highlighted while it plays.
• Jump to a paragraph: click its time.
• Edit a paragraph: hover over it, choose Edit, change the text, then Save. Word timings are kept where the words still match.
• Change a paragraph's speaker: choose a name from the menu at the start of the paragraph.
• Rename a speaker: in the Speakers tab, click the name, type a new one, and press Enter.
• Find in transcript: type a word. Matching paragraphs are outlined, and the box shows how many match.
• Clean view (in View): hides fillers and sound tags.
• Follow playback (in View): scrolls the transcript to the line being spoken.
• Undo all edits (in View): goes back to the original transcript.

Editing a paragraph turns off Clean view.

Export

1. In the Export tab, choose a Format:
   • Text (.txt)
   • Word (.docx)
   • Subtitles (.srt)
   • Subtitles (.vtt)
   • JSON with word timings
2. Tick Speaker names to include them.
3. For Text and Word, tick Timestamps to include them.
4. Choose Download.

With Clean view on, the export has no fillers or sound tags. Otherwise it exports exactly what's shown.

Limits

• Files up to 2 GB.
• Recordings up to 10 hours.
• Audio or video files: MP3, WAV, M4A, MP4, MOV, WebM and more.
• Up to 50 key terms, each up to 50 characters.

Credits

• 100 credits per minute of audio, rounded up. The minimum is 1 credit.
• The New transcript window shows an estimate once it knows the file's length.
• Credits are reserved from the file's length. If the speech heard runs longer, the difference is charged.
• If transcription fails, your credits are refunded.

Tips

• Set the language if you know it, especially for short clips.
• Add key terms for names and product words before you start.
• Turn off Label speakers for a single speaker.
• Use Clean view for readable notes, and the full view for exact quotes.
• Usually a transcript is ready in under a minute for an hour of audio.

Related

• Alignment: you already have the script and need exact timings.
• Dubbing: translate the recording.
• Clean Audio: remove background noise first.
• History: Open transcript from any page.