Workflow: Editing
A trim. A cut out of the middle. A fade. A LUFS normalize before upload. The Edit task runs jobs this size on the open file, from the keyboard or its task bar, keeps every edit in a history by name, and measures the file again after each one. Save As writes a new file; writing over the original is a setting.

The fix is thirty seconds of work. Opening a DAW to do it is not.
A few seconds of room tone off the head, a fade on the tail, a normalize to the LUFS target the delivery asks for: through a DAW, each means a project, an import, a bounce and a settings dialog. Done by hand on the master, there is nothing to fall back to if the gain is wrong.
The Edit task does the small jobs where the file already is. Every edit is undoable from its history, and Save As writes a new file, so the source stays as it was on disk. With Settings › Editing and plugins › Allow Save to overwrite the original file, Save (⌘S) writes over a working file instead, keeping its format, bit depth and channel layout, and swaps the new data in only once it is fully written.
Open the task, run the operation, save.
A peak normalize that only watches sample values watches the wrong ceiling. The number that gets a master rejected, and the one streaming platforms enforce, is the inter-sample peak: the level the signal reaches once the converter reconstructs the waveform between samples. A file can read −1.0 dBFS on the samples and still overshoot when it is played back.
Specula's limiter measures the 4× oversampled inter-sample peak and holds the ceiling against that, in dBTP. Normalizing in LUFS runs it whenever the gain raises the level, in two offline passes: the first builds the gain envelope from the oversampled peaks, the second catches any peak the release left fractionally over the ceiling.
Limit True Peak is the same limiter on its own, with no loudness measurement and no gain stage: for a file whose loudness is right and whose inter-sample peaks need capping under a platform's ceiling.
The Edit task has its own keys, active only there so they never shadow the transport, listed in the Edit menu and in Help › Keyboard Shortcuts. One group acts on the selection, the other at the playhead.
Level Dialogue is the edit for spoken word. It is a downward expander keyed to the speech regions Specula detects and the Speech task corrects: it ducks the room tone between phrases with a fast attack and a slow release, so word onsets and tails stay clean.
With Lift dialogue to a target on, it also brings the speech to a loudness target in the same pass, capping inter-sample peaks at a true-peak ceiling. The silence is gated first and the makeup gain goes on after, so the noise floor ends up lower, not higher, which is what makes it the fix for a room-tone floor sitting just over ACX's line.
It is a gate, not a denoiser. It rescues a marginal floor, not a hissy recording. Its main home is the Speech task's sidebar, where the Level dialogue panel predicts the dialogue loudness and the floor before Apply; the Edit task's Loudness menu opens the same edit. See the podcast and audiobook workflows for where it fits.
The Edit task hosts a chain of Audio Unit effects. The Plugins drawer adds any effect installed on the Mac (AUv2 and AUv3), chains several in order, bypasses or reorders them, and opens each plugin's own window, or a generic parameter panel for a plugin that ships none. It works on mono and stereo files, and the Edit menu reaches every drawer action.
The chain plays live while the file plays. Loudness details' Live side and the spectrum and phase panels follow the processed sound, while the delivery card and the whole-file readings keep describing the file until the chain is rendered.
Render bakes the chain into the audio through fresh, latency-compensated instances, as an edit in the history, and the file is measured again. With the effect tail off (the default) the rendered audio is exactly as long as the input, so it still lines up for an A/B and Align. Rendering clears the chain and remembers it, so Undo then Edit › Restore Last Plugin Chain is the tweak-again loop, and it survives quitting. Render to Compare Slot puts the processed version beside the original in Compare. AUv3 plugins run in their own process by default (Settings › Editing and plugins › Prefer out-of-process Audio Units), so a crash in one of them cannot take Specula down; AUv2 plugins run inside the app.

Spectral (⌘6) is for the blemish inside the audio worth keeping: a cough inside a piano chord, a chair squeak under speech, a bite of mains hum. Select it on a full-resolution, zoomable spectrogram, in time and frequency, then attenuate, erase or repair just that region, leaving the rest bit for bit. Include harmonics adds the overtones of a squeak or a whistle to the selection. It is a bigger job than the trims and fades on this page, so it has its own workflow: the selection tools, the wand's tolerance and reach, the feather, Include harmonics, and auditions before Apply.




The Edit task is for the jobs that do not earn a DAW session, most of them one key. It runs them on the open file, lists every edit in its history, and writes a new file on Save As, or over the original once that is allowed.