Stack audio adjustments and effects into one ordered chain. The same chain drives live playback and the final render, so what you hear while dialling it in is what the exported file contains.
Order is the signal path
Steps run top to bottom, each one processing what the previous step produced. That ordering is a real creative control, not bookkeeping:
- Echo before reverb sends discrete repeats into the room, so each repeat gets its own reflections.
- Reverb before distortion drives the reverb tail into the shaper, which smears and thickens it.
- EQ before distortion decides which frequencies get driven hardest, since the shaper responds to whatever arrives loudest.
- Fade last shapes the finished result, including any reverb tail that earlier steps produced.
Drag a step by its grip to move it, or focus the grip and use the arrow keys.
Building a chain that raises the level
A chain that raises level needs somewhere for the peaks to go. On most material the order that works is repair, tone, dynamics, ceiling:
- Cut before you lift. A high-pass at 30 Hz costs nothing audible and stops the shelf above it amplifying rumble into every step that follows.
- EQ before the exciter. The harmonic generator responds to whatever arrives in its band, so the EQ decides what it works on.
- Compressor before limiter. The compressor handles the average level over hundreds of milliseconds; the limiter catches the individual peaks the compressor let through.
- Limiter last. Anything after it can push the signal back over the ceiling it just enforced.
A fade is the one exception, because it shapes the finished result including whatever the limiter produced.
Bypass keeps the settings
The toggle on each card stops a step processing without removing it from the chain. Its settings survive, so comparing the chain with and without one step costs nothing and loses nothing.
Which steps can be chained
The chain processes audio as a stream, one block of samples at a time, and no step is allowed to see the whole file. That constraint is what keeps a 45 minute recording from being decoded into memory, and it decides what can be a step.
Level, tone, dynamics, and time-based steps all qualify: gain and EQ, the compressor and limiter, the gate and de-esser, the exciter and transient shaper, and every reverb, echo, and modulation effect each produce their output block from the current input plus their own internal state. Trimming and format conversion change the structure or length of the file, and loudness normalization needs to measure the whole thing before it can decide on a gain. Those remain separate tools with their own runners.
Presets and links
Save stores the current chain in this browser under a name, and re-saving under an existing name updates that entry rather than adding a duplicate. Presets are held in localStorage on one device.
Copy link encodes the whole chain into the URL, including every setting that differs from its default. That link is the portable form: it survives being pasted anywhere, and opening it rebuilds the chain exactly.
Rendering
Export runs the chain over the source through a background worker, writing the result to browser disk storage rather than holding it in memory, which is what makes long files workable. Reverb and echo tails are rendered past the end of the source so the decay completes.
Selecting a range on the waveform and turning on Export selected range renders only that section, which is the fast way to audition a chain on a chorus before committing to the whole track.