Online Audio Mixer - Layer Tracks & Export the Mix

Mixing is the craft of making several recordings sound like one performance, and the essentials have not changed since the first four-track machines: set each part's level, place it in the stereo field, decide when it enters, and listen to the sum. This mixer gives you those essentials in the browser. Each file you add becomes a clip on a shared multitrack timeline — its waveform drawn with wavesurfer.js, the open-source (BSD-3-Clause) waveform library — plus a track strip with a gain fader (0 to 150 percent), a pan control, and mute and solo buttons; a master fader scales the whole mix. Drag a clip along its lane to set when it starts, drag the dark strips at either edge to trim away a count-in or a ragged ending, and use the Zoom slider to stretch the timeline when you need to line a vocal up against a beat by eye. The Offset and Trim fields mirror every drag, so you can also type positions to the hundredth of a second. During playback the clips run through a live Web Audio graph — one buffer source per track, scheduled at its timeline offset and limited to its trimmed window, feeding a gain node and a stereo panner, all summed into the master. Click the ruler to move the playhead and audition any part of the arrangement; pause keeps your place, stop returns to zero. A few gain-staging basics go a long way: start every fader at 100 percent and pull tracks down rather than pushing one up, reach for pan before level when two parts fight for attention, and use solo to audition one layer in isolation — if any track is soloed, everything else is silenced, matching how hardware consoles behave. Keep the master at or below 100 percent, because summing several full-scale tracks can exceed digital full scale and 16-bit WAV hard-clips anything above it. The export is deliberately not a recording of what you hear. When you press export, the same node graph — offsets, trims, gains, pans, and the master level — is rebuilt inside an OfflineAudioContext, which renders the mix faster than real time with sample-accurate scheduling. The render is deterministic: the same clips in the same positions produce a bit-identical WAV every time, unaffected by CPU load, tab throttling, or dropouts during preview. The file runs until the last audible clip ends and is written as stereo 16-bit PCM at the highest sample rate among your tracks (at least 44,100 Hz). Typical uses: laying a voiceover under background music and sliding it to start after the intro, stacking harmony takes into one vocal comp, trimming and butting sound-effect beds against each other, combining separated stems back into a reference mix, or checking how a beat and an a-cappella line up before opening a full DAW. Because every stage — decode, preview, render — happens locally, unreleased music and confidential narration stay on your machine.