Punch-and-roll recording for voice over and audiobook narration

Read from the prompter, and when a line comes out wrong tap once: the punch point rolls back to the top of that line and you hear the real tail of your master as pre-roll. What you export is the assembled master — silences trimmed, seams crossfaded, real room tone in the gaps rather than the digital silence that gives a bad edit away.

Start See how it works

No account needed to start. Rehearse free for 7 days — 2 without an account. Studio — recording, the finished export and subtitles — is paid.

Studio
  • Arabic + English
  • Recordings never leave your device
  • Finished export

How it works for you

The record pass and the edit pass become one pass.

Step one

Load the script

Paste it or import a PDF or a Word file. It splits into breath-sized blocks and you set a words-per-minute target; a line like [0:30] on its own gives a section a target duration. The microphone opens once and the tape runs continuously, so the first word of a take is never clipped.

Step two

Punch and roll

One key kills the take and rolls the punch point back to the start of that same block, plays you the real tail of the finished master as pre-roll, and re-arms. Nothing after the punch point is destroyed, and a kill is reversible — alternate takes can be promoted later in Master.

Step three

Deliver

The master is assembled from your kept takes with per-take silence trimming, equal-power crossfades at every seam and real room tone in the gaps. Three presets: 16-bit mono WAV for AI voice tools, 192 kbps mono MP3 for sharing, and a small AAC in M4A that opens straight on an iPhone.

What a narrator gets

Only what the app actually does today.

  • Punch and roll without a DAW. Kill, roll back to the top of the block, hear your own tail, keep reading. It is the studio technique, run for you.
  • A master, not a pile of takes. Silences trimmed per take, equal-power crossfades at every seam, real room tone in the gaps — never digital silence.
  • Measured seams. On the current build the largest sample-to-sample step at a seam is smaller than the largest step in the untouched middle of a take (0.010 against 0.054), so no seam reads as a click.
  • Clipping is caught, not shipped. A take that hit full scale is flagged so you can re-record it, and it is never delivered at full scale.
  • Three export presets. 16-bit mono WAV (a clean source for AI voice tools), 192 kbps mono MP3 for sharing, and M4A for Apple and messaging.
  • Exact SRT and VTT with every export, timed from the edit rather than transcribed.
  • Nothing is uploaded. Client work stays on the machine you recorded it on — there is no server for audio, so there is nothing to leak and no NDA question to answer.
  • A crash buffer. Audio is written to your browser’s storage about once a second while you record, so a closed tab costs at most the last second. It is a buffer, not an archive: export when the session ends.

Questions narrators ask

Is this meant to replace my DAW?
It replaces the record-and-edit pass for spoken word, not the DAW. There is no multitrack, no music bed and no processing chain here — you get one clean assembled voice master, which you can deliver as it is or take into your DAW if the job needs more.
Can I still change my mind about a take?
Yes. A kill is reversible, nothing after the punch point is destroyed, and nothing you record is deleted without you asking. In Master you can play a block back, listen to an alternate take, and keep that one instead.
What about a bluetooth microphone?
Use a wired, USB or built-in microphone for anything you intend to deliver. iOS forces bluetooth microphones such as AirPods into a narrow-band call profile; the app warns you when it sees one.

One pass, and the master is done

Open the app and read a paragraph. Rehearse free for 7 days — 2 without an account; recording and export are Studio, which is paid.

Start

No account needed to start. Rehearse free for 7 days — 2 without an account. Studio — recording, the finished export and subtitles — is paid.