Guides › Recording

How to re-record one line so it matches the rest

A “punch-in” means recording just one line again and dropping it into the original take. Done well, nobody hears it. Done badly, the new line jumps out: closer, brighter, faster or flatter than everything around it. Here is how to get it right, before, during and after recording, and what to do when a punch-in can't match.

Can't record again in the same room? Fix the line in your own voice with Respeak →

Before you record: recreate the setup

• Same microphone, same position. Distance matters most: a few centimetres closer adds bass (the proximity effect) and makes the line sound “inside” the listener's head. Mark the mic position and your seat if you can.
• Same room and the same quiet. A different room changes the reverb and background noise, the most common giveaway. Turn off fans or heaters that weren't on before, or that were.
• Same input settings. Gain, interface, sample rate and any processing in your recording chain (noise reduction, compressor).
• Get back into the moment. Listen to the 30 seconds around the line a few times. Match the pace, energy and pitch you had, not how you would say it today.

While you record: give yourself a run-up

• Read the sentence before the one you are fixing as well, and carry on into the sentence after. You will only keep the middle, but the run-up brings back the natural intonation.
• Record several takes, then pick the one that fits best in context, not the one that sounds best on its own.
• Record 10 seconds of silence (room tone) in the same setup. You will need it to fill gaps.

After recording: edit it in

1. Cut at natural boundaries: in a pause, or just before a breath. Never in the middle of a word.
2. Crossfade both edges with short fades, about 10–30 ms for speech.
3. Match the loudness of the new line to the lines around it. Compare by ear and with a loudness meter.
4. Match the tone if the new line sounds thinner or boomier: a gentle EQ cut or boost, not a dramatic one.
5. Fill gaps with room tone so the background noise doesn't drop out around the edit.
6. Listen in context, 10 seconds before and after, on headphones and on a speaker.

When a punch-in can't match

Some situations make a matching punch-in almost impossible: the recording is weeks old, it was made somewhere else or with a different mic, your voice has changed (a cold, a long day), or the original was a live conversation with energy you can't recreate alone. Two ways out:

• Regenerate the line in your own voice. Respeak transcribes your recording; you edit the sentence as text, and it's read in a voice made from that recording. Tone and room sound are matched to the audio around it, so there's no setup to recreate. You get 3 free generations per sentence before replacing it (1 credit).
• Record the line in the editor. If you'd rather say it yourself, Respeak also lets you record that one sentence in the browser and drop it into place. That doesn't use a credit.

Try it on your recording. 10 free credits when you sign up →

Limits: regenerating only works with your own voice, in Chinese (Mandarin), English, Japanese, Korean, Spanish, French, German and Portuguese, and it works best on clean speech. Loud background noise or music under the voice makes it harder.

FAQ

Why does my re-recorded line sound different?

Usually because something changed: the distance to the microphone (closer sounds bassier), the room, the input gain, or your voice and energy on a different day. Match those first; fixing it with EQ afterwards is harder.

How long should the crossfade be?

For speech, short crossfades of about 10–30 ms at each edit point usually hide the cut. Cut in a pause or just before a breath, not in the middle of a word.

What if I can't record in the same place again?

Then a punch-in will rarely match. Regenerating the sentence in your own voice from the original recording avoids the problem, because the voice and room come from the recording itself.