Do both guests need to be on separate channels?

Yes, and the tool enforces it. It requires exactly two channels and rejects mono, so the real precondition is a recording where each microphone landed on its own side. If both voices share one channel, or both microphones picked up both voices, this tool cannot separate them.

What does the target level mean?

It is the block RMS level that each channel is steered toward, expressed in dBFS, with the slider covering -24 to -12 dBFS and a default of -16. The leveler compares each 100 millisecond block against that target and requests the gain needed to reach it, subject to the max-gain and slew caps.

Why is the max speed setting important?

Because it is what stops the leveler sounding like it is pumping. The slew limit caps how far gain can move per block, so at 18 dB/s the gain moves at most 1.8 dB every 100 milliseconds. Turn it up for faster correction on long level changes and down for a gentler, less obvious ride.

Will the report tell me what gain was applied?

Yes. After processing, a readout lists each guest's measured input level, the resulting output level, and the range of gain applied as a low and high figure in decibels. That range is useful for seeing whether a channel was riding the max-gain cap, which is a sign the microphone was too far away.

Does it fix a guest who sounds thin or noisy?

No. This is a gain tool and nothing more. It balances loudness, and a channel that is thin, boomy or hissy stays that way after leveling. Tone and noise belong to a different tool, and lifting a hissy channel with gain can make the noise more audible, which is why the max-gain cap exists.

Multi-Guest Leveler

You Might Also Need

Audio Leveler

Automatically level audio volume

Loudness Normalizer

Normalize to LUFS loudness standards

RMS Normalizer

Normalize audio using RMS levels

Audio Compressor

Dynamic range compression for audio

Multi-Guest Leveler balances two speakers who were recorded onto separate channels of one stereo recording, steering each channel to the same target level on its own. It is built for the standard two-microphone setup where one guest sits left and the other sits right.

What is Multi-Guest Leveler?

Drop a stereo recording in which each guest occupies one channel and the tool levels both independently to a common target, then hands back a balanced WAV. The assumption is explicit: this works when the microphones were kept apart, so the left channel is essentially one voice and the right channel another. Two guests recorded that way, one quieter and one louder, come out matched without touching the other channel, and the tool reports the measured before and after levels plus the gain range it applied to each side.

The three controls, with their real ranges:

Target level per guest, adjustable from -24 to -12 dBFS RMS and defaulting to -16 dBFS.
Max gain per guest, from +6 to +24 dB and defaulting to +15 dB, which caps amplification on a near-silent channel.
Max speed, from 6 to 40 dB per second in steps of 2 and defaulting to 18 dB/s, which limits how fast gain is allowed to move.
A required stereo input: the tool rejects mono and anything with more than two channels and says so rather than guessing.
A readout after processing showing each guest's input level, output level and the gain range applied.

The leveling itself is a block-based automatic gain routine run once per channel. The signal is cut into 100 millisecond blocks, each block's root mean square level is measured, and the gain needed to reach the target is computed in decibels. That request is clamped so it never exceeds the max gain per guest, then rate-limited so the change from the previous block is at most the slew allowance, which at the default of 18 dB/s works out to 1.8 dB per block. The resulting gain is applied to the whole block and the next one is evaluated from the new position. Because the slew limit is applied after the target is clamped, gain eases toward the target instead of jumping, which is what keeps a syllable from pumping. The minimum and maximum gain reached during the pass are tracked and reported back as the gain range.

The limits worth knowing before you rely on it:

It needs exactly two channels, so a mono mixdown or a single-microphone recording cannot be split by this tool.
It cannot separate guests that bled into both microphones; if both voices are audible on both channels, leveling a channel levels both voices together.
It corrects level only, with no EQ, de-essing or noise reduction, so a thin or noisy channel stays thin or noisy, just louder.
The max gain cap means a genuinely silent channel is left quiet rather than being lifted into audibility.
Output is a 16-bit WAV, so a long episode becomes a large file even though the input was compressed.
The report describes levels and gain range only; it is not a waveform view and offers no per-word correction.

Use it immediately after a two-microphone capture, before any other processing, and let slow slew keep the movement invisible: 18 dB/s is a good default because it reacts within a second or two without chasing every syllable. Raise the max-gain cap only when a guest was recorded genuinely quiet, and lower the target if the balanced result sounds loud against your other episodes. If both speakers are already on the same channel, this is the wrong tool; a general leveler or a per-channel edit is the right one.

How to use Multi-Guest Leveler

Drop a stereo recording where each guest is on their own channel, left for guest one and right for guest two.
Set the target level per guest, between -24 and -12 dBFS RMS, with -16 dBFS as the usual starting point.
Adjust the max gain per guest and the max speed if you need more headroom or faster correction.
Press Process, check the reported gain range per guest, then preview the result and download the balanced WAV.

When to use Multi-Guest Leveler vs related tools

Multi-Guest Leveler is the two-channel, two-person case, and Audio Leveler is the general one that levels a whole file without assuming anything about channel layout. When the destination is a streaming platform with a published loudness target, Loudness Normalizer matches an integrated LUFS figure instead of block RMS, and RMS Normalizer does the same job against RMS. If the problem is peaks rather than average level, Audio Compressor shapes the dynamics with a threshold and ratio.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.