AetherCutChangelogHow it worksPricingOpen the editor

Home / Changelog / v1.5.9

v1.5.9

AI Voiceover — now audible in preview and baked into exports

AetherCut v1.5.9 shipped on Feb 2026.

What's new in v1.5.9?

  • Hidden <audio> element in the preview pipeline syncs voiceover playback with the main timeline (play/pause/seek, drift-corrected)
  • Voiceover volume from the Audio panel now controls live playback in real time.
  • Export pipeline taps the voiceover AudioContext graph via MediaStreamDestination — voiceover is mixed into the exported WebM/MP4.
  • When a voiceover is present, export automatically uses the MediaRecorder path (WebCodecs can't capture audio elements); pure-video exports keep the faster WebCodecs path.
  • Known gap: clip-native audio + bgMusic still not included in export — follow-up work.

Making the most of AI voiceover in preview and export

If you are opening v1.5.9 and wondering what to test first, the simplest proof is a short scene with a single AI voiceover line and one obvious visual beat. Add a captioned intro card, place the playhead at the start, and preview from the canvas instead of scrubbing frame by frame. You should hear the voice begin with the timeline, stop instantly when you pause, and resume in the right place after a seek. That sounds basic, but it changes the editing feel in a real way: you can now judge pacing, line breaks, and visual timing while hearing the same generated narration that will be carried into the rendered file. Before this release, it was too easy to build cuts against a silent mental model of the voice, then discover during export or external review that the line landed early, stepped on a transition, or left dead air after the last word.

A practical way to work with this is to treat the voiceover as your timing spine before you polish motion. Start by dropping in the AI narration, then use natural speech landmarks to place visuals: the first noun can introduce the subject on screen, a short pause can hold a logo or product shot, and the final stressed phrase can line up with your CTA or end card. Once that rhythm feels right in preview, open the Audio panel and adjust volume while the timeline is playing. Because the level change now affects live playback, you can make decisions by ear instead of editing blind and hoping the export matches. This is especially useful when the voice needs to sit above music conceptually, even though background music export is still a known gap. You can still tune the narrator’s level now so that when the rest of the export audio path lands in a follow-up release, your voice track is already behaving the way you want.

There are a few easy mistakes to avoid when testing the new behavior. The first is assuming every export path behaves the same. In v1.5.9, the app switches export strategy when a voiceover exists, because the faster pure-video route cannot capture audio elements. That is expected, not a regression, so a voiceover project may export differently from a silent motion piece. The second is forgetting what is and is not included yet. If your sequence also uses clip-native audio or background music, hearing the AI narration in preview can create the impression that the exported file will contain the full mix. Right now, the useful mental model is narrower: the generated voiceover is the piece that is now audible during editing and baked into the final WebM or MP4. If you need a final deliverable with other audio sources today, check the output file immediately instead of assuming parity with the preview. The third mistake is over-correcting volume after a single loud phrase. Because you can now hear the adjustments live, it is tempting to chase every moment. A better approach is to listen to the full sentence or the full scene and set a level that works across the passage, especially if your visuals rely on steady perceived loudness.

For anyone using AI voiceover as a fast drafting tool, this release also makes iterative scripting less clumsy. You can generate a line, preview it against the timeline, and decide whether the problem is the wording or the edit. If the sentence feels rushed, you may not need to slow animation; trimming adjectives or splitting one long thought into two shorter lines often solves the timing more cleanly. If the narration sounds fine but the scene drags, tighten the visuals around the spoken beats rather than rewriting a line that already works. In other words, preview-with-audio helps you separate writing issues from sequencing issues earlier. That matters most on short-form explainers, product demos, and social clips where a half-second mismatch is noticeable. The big win in v1.5.9 is not only that AI voiceover now survives export, but that your decisions while editing can finally be based on what viewers will actually hear.

Which features does this release touch?

How do I get this update?

Open the editor and you have it. AetherCut runs in Chrome, Edge, Safari and Firefox. There is nothing to install, and you can edit without an account. Open the editor or see how it works.

What shipped just before and after?

See all 111 releases · Follow the RSS feed