Release video
What's new in edytlab v0.7.0
A 58-second video of what's new in edytlab v0.7.0: a typed edit, steady faders, the history graph and a 200-track session. Real app, scripted assistant.
0:58, with captions you can switch on. Nothing plays until you press play.
What it shows
What's new in v0.7.0, shown in the real app. It starts with a request typed in plain English that the assistant turns into a fade and a gain change, which you then hear. Then six changes: faders that stay where you put them (#596); the history graph keeping up as a version is named and the history compacted (#602); a 200-track session where list_session pages through what the summary leaves out and twelve string tracks come down 4 dB (#618); the ONNX Runtime now shipping in every installer, as groundwork only, with no model yet; bundled skills that update unless you edited them (#574); and playback that survives switching tabs (#607). Real app, scripted assistant: every shot is the real v0.7.0 frontend, the assistant's replies and tool calls are scripted, and the demo audio is made in code, as the note on every shot says. The installers are unsigned. Captions are available.
What happens, in order
There is no voice-over. Each line is a moment in the video, from the second it starts.
- 0:00Title: edytlab v0.7.0, “Describe the edit. It happens.” Every app shot carries the note “v0.7.0 · real app · scripted assistant”.
- 0:03Describe the edit in plain English: the request is typed into the chat box (shown at 1.5× speed).
- 0:06The assistant calls the tools, fade and then set_track_gain, and the timeline updates: the backing fades in over 4 seconds and the vocal lane reads −3.0 dB.
- 0:12Press play and hear it; the demo audio is made in code.
- 0:16Faders that stay where you put them (#596): four key taps on a gain slider step it +0.5, +1.0, +1.5 and +2.0 dB, with no bounce back.
- 0:20Every edit is a node in the history graph.
- 0:23The history graph keeps up (#602): naming a version shows at once, and compacting the history removes nodes at once, 17 nodes becoming 8.
- 0:28200 tracks? Just ask (#618): list_session pages through the session (shown at 2.3× speed), then twelve string tracks sit at −4.0 dB.
- 0:37The ONNX Runtime now ships in every installer: groundwork only, no model yet.
- 0:41Bundled skills update when a release changes them, unless you edited them (#574).
- 0:45Switch tabs, press play: it still plays, with no error (#607).
- 0:52End card: edytlab v0.7.0, edytlab.com, free and open source (MIT), for macOS, Windows and Linux. The installers are unsigned.
More videos
- Launch video: a DJ job from request to export
- Beatmatch and blend two tracks
- Extend an intro for mixing
- A three-track mini-mix
- What's new in v0.7.0 (short reel)
- Describe the edit, and it happens (short reel)
- 200 tracks? Just ask (short reel)
- Highlights: three DJ jobs in a minute
- Teaser: ask for a blend, hear the result
- Beatmatch and blend two tracks (short reel)
- Extend an intro for mixing (short reel)
- A three-track mini-mix (short reel)
- The AI audio editor for DJs page, where the three full demos also play
edytlab is free and open source (MIT) for macOS, Windows and Linux. Stem separation and transcription are not built yet, and the current builds are unsigned, so macOS and Windows warn on first launch; the setup guide has the steps. Download it from the releases page.