Problem-solving

Why Do Merged Audio Files Change Volume Between Clips? A Measured Test

Merging audio into one output format does not automatically equalize its volume. We measured a 12 dB difference across two synthetic clips merged with Toolit's actual processing arguments.

When the second part of a merged meeting recording suddenly sounds louder, did merging fail? Making clips compatible for one output file and adjusting how loud each segment sounds are different operations.

We joined two signals with different levels using Toolit's actual audio-merging arguments. We measured the resulting segments and explain when to adjust source levels before joining them.

Two different kinds of matching

Matching format does not equalize volume

Toolit converts inputs to 44.1kHz stereo before concatenation. Its current merge path has no per-file volume or loudness matching stage.

The 12 dB test gap remained

Two two-second synthetic signals with a fourfold amplitude difference measured -36.1 and -24.1dBFS in the joined output.

Consistent listening may require editing first

Listen to each input before merging. Peak matching and matching average perceived loudness are different jobs.

Two equally long signals at different levels

  1. 1

    Using Toolit's bundled FFmpeg WebAssembly engine, we generated two two-second pure tones at 440Hz and 660Hz. The second source had four times the signal amplitude of the first. These are convenient measurement inputs, not a voice or music listening test.

  2. 2

    We merged the WAV sources with the tool's actual argument builder into one 44.1kHz stereo WAV. We calculated RMS signal level in dBFS over 0.2–1.8 seconds and 2.2–3.8 seconds of the result, excluding the boundaries.

  3. 3

    The reproducible script is scripts/measure-audio-merge-levels.mjs. dBFS describes a digital signal measurement here; it is not a LUFS perceived-loudness measurement or a human listening result.

The difference survived concatenation

The output became one four-second, 44.1kHz stereo file. Yet the measured gap between its first and second segments remained 12.0dB, as expected from the input amplitude ratio. Matching file specifications is not the same as matching volume.

Output segmentTest inputMeasured RMS
First 0.2–1.8 secQuieter 440Hz tone-36.1dBFS
Second 2.2–3.8 sec660Hz tone at 4× amplitude-24.1dBFS
DifferenceOne output format12.0dB

Same format is not the same level

Inputs with different sample rates or channel layouts must be made compatible for one output stream. Toolit's merge command does this and then concatenates them. It has no filter that measures and corrects the amplitude or perceived loudness of each file.

So support for joining different input formats does not mean every part will sound equally loud. Differences in how the source recordings were made can persist in the output.

Would matching peaks be enough?

Not always. A recording with one brief loud sound and otherwise quiet speech may have the same highest peak as another recording while sounding quieter for most of its duration. Audacity distinguishes peak-based normalization from perceived-loudness normalization for this reason.

Our pure tones do not represent speech, music, background noise, or silence. A dBFS gap in these signals is not a direct claim about how loudly a person would hear two real recordings.

What should you check before merging?

Listen to the beginning, middle, and end of every source with the same player volume. If a jump would bother the listener, adjust levels in an audio editor that supports that job. Then listen again for distortion or noise before joining the prepared clips.

If the sources already sound consistent and you just need one ordered file, extra adjustment may be unnecessary. For music, forcing every piece to one numerical level can undermine intentional dynamics.

SituationUseful next step
Meeting clips vary in speech levelListen and adjust individual files when needed before merging
Clips already sound consistentMerge, then check joins and full playback
Musical dynamics matterDo not flatten levels solely to match a number

What this test cannot claim

We deliberately set a 12dB difference between two simple signals and tested WAV output. It does not mean all merged files differ by 12dB. We did not test MP3 or AAC encoding error or measure perceived loudness in people. Inspect the joins in your own material.

Official documentation used for this analysis

Merging joins clips; matching loudness is separate

Putting audio into one output format does not remove the original volume difference. Our four-second merged file retained a measured 12.0dB gap between segments.

Listen to the sources and final output at the same playback volume. If the transition is distracting, adjust source levels first. Toolit's current audio merger does not automatically equalize loudness.