Skip to content
PaperZero

Audio Transcript to PDF

Open a browser-decodable audio recording without uploading it, listen in the local player, paste or type a reviewed transcript, insert timestamps from the real playback position, and export a paginated PDF. Automatic speech recognition is deliberately not bundled in this build and no cloud speech service is contacted.

  • Processed locally in your browser
  • No watermark
  • No sign-up
  • Works offline

Conservative local fallback: this build does not bundle a roughly 100 MB on-device speech model. It never sends audio to a browser or cloud speech API. Listen locally, paste or type a reviewed transcript, and optionally insert timestamps from the player.

Choose an audio recording

MP3/WAV/M4A or browser-decodable audio · one file · maximum 250 MB

Separate paragraphs with a blank line. Timestamps use [MM:SS] or [HH:MM:SS].

Compatibility report

PaperZero reports what this browser-local renderer can reproduce; it does not claim pixel-perfect source pagination.

Preserved

  • editable transcript text
  • optional title and date
  • user-inserted timestamps
  • paragraph breaks

Omitted

  • audio media embedding
  • automatic speech recognition in this build
  • speaker diarization
Transcript PDF layout

How it works

  1. 1Choose an MP3, WAV, M4A, or other browser-decodable recording; it stays in the current tab.
  2. 2Listen with the local media player and paste or type a transcript you have reviewed.
  3. 3Optionally insert the current playback time as a real transcript timestamp and add title/date fields.
  4. 4The shared worker paginator creates and validates the transcript PDF without embedding or uploading the audio.

Your privacy

This tool runs entirely inside your browser using JavaScript and Web Workers. Your document is never uploaded to any server — you can even disconnect from the internet and keep working once the page has loaded.

Frequently asked questions

▸Does PaperZero automatically transcribe my audio?

Not in this build. A suitable on-device Whisper package would add roughly 100 MB and substantial device requirements, so this route provides the phase's conservative editable fallback and says so before file selection.

▸Is my audio sent to a speech API?

No. The recording is played from a temporary browser object URL. No browser or cloud speech-recognition service is called.

▸Are speakers identified automatically?

No. Speaker labels are not claimed because diarization is not implemented. You can type labels into the reviewed transcript yourself.

Related tools

Donate