Audio Repair and Restoration

Two people talked over each other on one mic: what you can fix in REAPER, and what needs a separation tool

By StudioRescue Editorial

Published October 31, 2026

You play back the interview and there it is: the best answer of the episode, with your guest’s co-host laughing and agreeing right over the top of it. Or two people argue a point for ten seconds and you cannot follow either of them. Everything was recorded on one mic, or one phone, or one camera, so there is only one track.

This is a different problem from mic bleed. With two mics, each person has a clean track and the job is to silence the other one, which is covered in how to fix mic bleed between two podcast mics. With one mic there is nothing to switch to. Here is what is realistic.

Why EQ, a gate and noise reduction cannot separate them

Each of the usual repair tools needs the unwanted sound to be different from the wanted sound in some simple way.

• EQ works when the unwanted sound sits in different frequencies. Two speaking voices overlap heavily in frequency, so any cut that removes one person removes most of the other too.
• A gate works on the level of the whole track. When both people are talking, the track is loud, so the gate stays open for both.
• Noise reduction learns a steady background sound and subtracts it. A second voice is not steady, it is speech, exactly like the voice you want to keep.
• Spectral editing in REAPER lets you change selected frequencies inside a selected stretch of time. The official user guide gives breathing sounds and sibilance as examples. It works on a sound that occupies its own small corner of the spectrogram, not on a second voice laid across the same frequencies at the same moment.

Step 1: find and sort every overlap

Listen through once and drop a marker at each overlap (press M at the play cursor). Then sort each one into one of three kinds, because each kind has a different answer.

Descript’s editing guide on crosstalk makes a point worth keeping in mind here: some overlap is a sign of real conversation and real feeling, and removing every trace of it can make people sound stilted. Cut what is hard to follow, not every overlap you can find.

Step 2: cut and smooth the overlaps in REAPER

On a single-track recording this is ordinary editing, and it is safe because there is no second mic track to keep in sync.

1. Turn on automatic crossfades so every cut gets a short fade instead of a click. The user guide covers the defaults under Options > Preferences > Project > Item Fade Defaults.
2. Choose your ripple mode. Alt+P cycles ripple editing between off, per track and all tracks. Per track is convenient when there is only the one voice track. If you have music or an intro on other tracks, make sure the mode you pick will not shift them out of place. The details are in how to cut a long podcast interview down to length.
3. Split at the edges of the overlap with S and delete the shared stretch, or move one edge to trim a speaker’s start. Cut in a breath or a gap where you can, not in the middle of a vowel.
4. Listen across every cut at normal speed. If a cut sounds abrupt, lengthen the crossfade a little or move the cut point into the nearest breath. The same splitting and fading mechanics are shown step by step in how to edit out filler words and breaths in REAPER.

For a loud laugh or exclamation that you decide to keep, you can split around it and pull that piece down with the item volume handle (drag the top edge of the item). Be clear about what that does: it lowers both voices for that moment, because they are in the same audio. It makes the moment less jarring. It does not remove anyone.

Step 3: when a separation tool is worth trying

Until recently the honest answer was that one-mic overlap could only be kept or cut. That has started to change. Several tools now say they can split two overlapping voices from a single recording. StudioRescue has not tested any of them. What follows is what their makers and one hands-on review say, so treat each as a claim to check on your own audio.

Similar voices are the hard case. In a 2024 Adobe community thread about two co-hosts whose voices sounded alike, an Adobe community expert said separating them was not possible in Audition and that similar voices would challenge even an advanced AI system. The newer tools above came out after that thread, but their own stated limits point the same way.

If you try one, keep REAPER as the home of the edit:

1. Make a time selection around the overlap with a second or two either side, and render just that stretch (set the render bounds to Time selection, covered in how to export audio from REAPER). Short clips keep cost and processing time down.
2. Run the separation in the other program and export each speaker as its own file.
3. Bring both files back into REAPER on two new tracks, lined up exactly with the original stretch, and crossfade them in over the original.
4. Solo each separated track and listen for warbling, metallic tails or leftover words from the other person. If the result is worse to listen to than the overlap, go back to cutting.

What this will not do

• It will not separate two voices with stock REAPER tools. Neither the current REAPER user guide nor the Cockos changelog describes a speaker-separation feature.
• Cutting will not keep both people’s words from an overlap. Every cut through a shared stretch loses some of someone.
• A separation tool will not guarantee clean results, especially with similar voices, three or more people, loud background noise or long stretches of overlap.
• None of this will change the room sound or distance. If one person was far from the mic, a separated track of them will still sound far away.

Stop it at the source next time

• Give each person their own mic on their own track. This is the change that makes overlaps editable at all. The setup is in how to record two people in one REAPER session. You trade this problem for mild bleed, which is far easier to fix.
• For calls, record each side separately. Zoom’s desktop app, for example, has a setting to record a separate audio file for each participant, and the options for other setups are in how to record a Zoom or phone interview into REAPER.
• Agree a simple signal before you start. A raised hand to say “I want to come in” costs nothing and saves editing time.
• Leave a beat before answering. Half a second of gap between speakers gives you a clean place to cut.