Stack music, voiceover and effects on audio lanes, duck the bed under speech automatically, cut the silences and filler words in one pass, and clean up noisy recordings at export.
Lanes, fades, ducking, keyframes
Audio lives on stacked lanes with waveforms, per-clip volume and edge fades. Flag the music bed to duck and it dips under any voice automatically; volume keyframes handle anything custom. Two scans tighten a talking-head recording: one finds the silences, the other finds the filler words, and you approve the cuts from a checklist.
Silence removal: one scan, a checklist, one click of jump-cuts
Volume keyframes for anything the automatic passes do not cover
Per-clip volume and edge fades, so any cut can carry a crossfade
Generate sound, then clean it up
Generate music, natural voiceovers, sound effects and clip-matched foley from a prompt. Shape the result with a 3-band EQ, pitch shift, voice presets, echo and room reverb, all previewed live. Export-time cleanup normalises loudness and reduces noise on rough recordings.
AI music, TTS voiceover, SFX, and video-matched foley
Pitch shift and voice presets alongside the EQ
Record in the browser; cleanup applies at render time
Rescue a recording, or take a mix apart
Feed a track you already have to a model and get a changed one back. Clean up a noisy voiceover, lift the vocal out of a song, stretch a thirty-second loop to fill the timeline, or regenerate one bad bar in the middle.
Denoise a rough recording, or isolate the voice from its background
Split a song into stems: vocals, drums, bass, guitar or piano
Extend a track to fit the cut, patch a section, or swap the voice
Questions
Can it duck music under my voiceover automatically?
Yes. Mark the music as Duck and the mix dips it whenever the voice lanes carry speech, in the preview and in the export.
Can I generate a soundtrack?
Yes: music models (with lyrics support), ElevenLabs voiceovers, sound effects, and foley matched to a video clip's content.
Can I clean up a noisy recording?
Two ways. Export-time cleanup normalises loudness and reduces noise on any clip for free. For something rougher, the rework models denoise the track or isolate the voice from its background, for a fraction of a cent at editor lengths.
Can I get the vocals out of a song on their own?
Yes. Stem splitting returns the part you ask for (vocals, drums, bass, guitar or piano) as its own audio file, and a prompt-driven separator can isolate anything you describe.
My music track is too short for the edit. Can I extend it?
Yes. The extend model continues the track in the same style, adding up to a minute at the end or before the start, instead of looping it audibly.
Can I record narration?
Yes. An in-browser voiceover recorder captures your narration and drops it straight onto an audio lane.
Can it remove silences and filler words?
Yes. Silence removal scans the clip and proposes jump-cuts for every quiet stretch; filler-word removal transcribes the speech and finds the "um"s and "uh"s. Both show a checklist: untick anything you want to keep, cut the rest in one click, and undo with one Cmd/Ctrl+Z.
What cleanup and effects are available?
Loudness normalisation and noise reduction per clip at export. EQ, pitch shift, voice presets, echo and room reverb preview live and render the same way.