Audacity clipping: check the recording before making it louder

Identify where the problem starts

A voice recording that sounds quiet and one that sounds distorted need different investigations. Begin with a preserved source file and a short passage that includes the loudest speech. Listen at a comfortable playback volume before adding processing. This procedure is an editorial workflow based on Audacity documentation, not a measured microphone comparison or audio-restoration test.

Audacity's test-recording tutorial describes clipping as a recording level that exceeds what can be recorded, leaving truncated waveform peaks. Its Normalize documentation describes changing peak amplitude and optionally removing DC offset. These are different operations: an amplitude adjustment is not evidence that damaged source information has been reconstructed. [1][2]

Make a test take before the full narration

Use the intended microphone position, input device, and ordinary speaking style. Include a naturally louder phrase rather than testing only a whisper. Record the settings that you can identify without assuming that the operating-system control and the microphone's hardware gain are interchangeable. Keep the recording short enough to inspect carefully.

Audacity recommends inspecting a test recording and lowering the recording level before trying again when clipping occurs. It also documents a clipping display in the waveform view. Use the visual indication together with listening; the purpose is to find a problem before committing a long narration, not to chase a particular waveform appearance. [1]

Separate capture, editing, and playback

If the unprocessed source already sounds harsh on the loud passage, compare a new take at a lower input level and consistent microphone distance. If only the edited mix distorts, inspect processing and combined track levels instead. If distortion appears only on one playback device, include that distinction in the record rather than discarding the source immediately.

Preserve an untouched original and name the processing copy clearly. Change one variable at a time so the comparison answers a question. Do not apply several enhancement effects together and call the last setting the cause of improvement. The worksheet records source observations separately from processed-export observations.

Worked example: a fictional voice-over

In this simulated example, a narrator's quiet opening is followed by a loud sentence with visibly truncated peaks and audible harshness. Applying a lower normalization target makes the whole clip quieter, but the harsh sentence still sounds damaged. Because the narrator is available, the editor records a replacement passage at a lower input level.

The replacement is reviewed in the surrounding edit for tone, timing, and background consistency. The editor keeps the original take and documents the replacement rather than claiming a repair algorithm recovered it. This is an invented decision example; it does not establish a universal input setting, measured distortion threshold, or guaranteed restoration method.

Normalize with a defined purpose

Audacity's Normalize effect can set selected tracks to a common peak level. It also offers stereo-channel handling choices, so inspect the selection and options before applying it. Equal peak values should not be treated as an editorial judgment that two speakers sound equally loud. Listen to transitions in the actual mix. [2]

Choose the final export requirements from the real destination or client brief; this article does not invent a universal loudness target. Play the exported file, including the loudest phrase and track transitions, outside the editor. Mark source quality, processing, and delivery separately. When rerecording is impossible, document remaining distortion and obtain the appropriate editorial decision rather than labeling the audio fully restored.

Sources & verification

Product details and prices can change. Check the linked provider before buying.

1. Audacity: Tutorial - Making a Test Recording Accessed 2026-09-14
2. Audacity: Normalize Accessed 2026-09-14