Waveform & Peak Viewer

Overview

Most waveform views scale every file up to the loudest peak, so everything looks the same. This one doesn't. The vertical axis is full scale, so a master 8 dB quieter looks 8 dB quieter. Flat tops where samples sit at full scale are marked as clipping, and true peak is measured at four times the sample rate to catch the peaks between samples.

Check a master for clippingSee at a glance whether a limiter was pushed into flat tops.
Compare two versionsOpen a mix and its master and compare their real levels.
Find the loudest momentThe true peak is marked where it happens, so you can go straight to it.

Set

Audio Tools 17

Neighbours

01 Audio Transcriber 02 Generate AI Music 03 Audio Converter 04 BPM Detector 05 DJ Booth 06 Extract Audio from Video

Specification

Command: audio.waveform_viewer
Set: Audio
Input: MIXED
Output: —
Max file: 1GB
Batch: single file

MPEG WAV FLAC MP4 OGG OPUS AAC WEBM

Runtime

Engine: ffmpeg-wasm 0.12
Threads: — workers
Isolated: —
WebGPU: —
Memory: —

Transport 0 B

Your file is opened, worked on and saved inside this tab. This tool has no upload step, so your file has nowhere else to go.

Keys

Search: ⌘K
Panels: ⌘1 · ⌘2
Full width: F
All keys: ⌘/

How it works

Drop a file

It's decoded on your computer and never uploaded.

Read the shape

The waveform is drawn at full scale, so quiet looks quiet and loud reaches the edges.

Look for the marks

Clipped runs show as bands at the top and bottom, with true peak and sample peak printed below.

Frequently asked

Why is the waveform not normalised?

Because a normalised waveform can't be compared with anything. Stretching every file to fill the box hides the level difference you want to see, say between a mix and its master, or two takes.

What counts as clipping here?

A run of samples in a row at or above full scale. A single sample at 0 dBFS is normal; a flat top of several means the waveform was cut off. Only runs are marked.

Why is the true peak higher than the sample peak?

Because the real waveform passes between the samples. Rebuilt at four times the rate, a signal whose samples never pass −3 dBFS can still reach 0 dBTP in between. That in-between level is what clips converters and lossy encoders, so it's shown next to the sample peak.

Is my audio uploaded?

No. Everything happens in your browser.

Keys

Search: ⌘K
Panels: ⌘1 · ⌘2
Full width: F
All keys: ⌘/