Timeline Studio

Timeline Studio is a video editor that runs entirely in a browser tab, with the AI parts running on your own machine rather than on somebody's GPU cluster. It handles multi-track timeline editing, generates voiceovers from text, transcribes audio into captions, animates a still portrait into a talking avatar, and exports finished video as MP4 or WebM. There's no account, no upload step, and no payment screen anywhere in it.

The interesting engineering choice is that inference happens locally through ONNX models the browser downloads and caches. Text to speech runs on Kokoro-82M, transcription runs on Whisper, and the talking avatar work runs on Live-Portrait, with the models pulled from HuggingFace on first use and stored in browser storage afterward. Execution goes through WebGPU where the hardware supports it and falls back to WebAssembly where it doesn't. There's a settings panel for checking browser support and clearing the model cache when it fills up, which you'll want, because these models aren't small and the first run of any given feature involves a real download. After that they're cached, so the cost is paid once per feature rather than once per generation.

The practical payoff is privacy and cost at the same time. Your footage never leaves the machine, because there's no server to send it to, and that's also why the tool can be free without a credit card or a usage meter. Anyone who has hit a monthly cap on a hosted AI video tool, or hesitated before uploading client footage to a service whose data policy they hadn't read, will recognize what that buys. The same architecture means it works offline once the models are cached. It's a genuinely different bargain from the hosted tools. Rather than paying a subscription so that somebody else's GPU does the work, you spend one download and then your own hardware does it indefinitely, for nothing.

Feature coverage is broader than the browser-based framing suggests. Beyond the timeline, there's AI vocal separation for pulling voice off a backing track, waveform analysis, audio reverse, background music, recording straight to a voiceover track, caption styling with lip-sync alignment, effects, stickers, transitions, and vision analysis for a clip or a still. The interface ships in ten languages including English, Chinese, Japanese, Korean, Spanish, French, German, Portuguese, Thai, and Vietnamese. It's installable as a progressive web app with a standalone window and a shortcut that opens straight into voiceover editing. That breadth is itself a fair signal of intent, since a thrown-together shell would have stopped at one headline feature rather than shipping vocal separation and lip-sync alignment.

Two clarifications matter before anyone tries it, and both come from checking rather than from the marketing. The first is the name. There's an unrelated project also called Timeline Studio from chatman-media, which is a Tauri desktop application distributed through GitHub under MIT with a Commons Clause. This is not that. This one is a browser app with no public repository and no license statement, so despite what some listings say, it shouldn't be treated as open source. It's free, which is a different thing. The second is the address, since it lives on a subdomain of ai-creator.top, a hub of free browser-based AI image and audio tools. The throwaway-looking domain undersells what's actually there, because the app itself is substantial and genuinely works. The bundle is over a megabyte of real application code wired to real models, which isn't what a scraped landing page looks like underneath.

The honest limits follow from the design. Local inference means your laptop is doing the work, so quality and speed depend on your hardware, and a machine without WebGPU will fall back to WASM and feel it. Browser-side export won't match a native editor on long or complex timelines, and model downloads make the first run of each feature slow. There's no cloud project storage, so work lives in the browser session, and generated voiceovers are saved there rather than in an account you can log into from another device. This isn't a replacement for a professional NLE on a feature film.

For its actual audience, though, the trade lands well. Someone making short social video, a voiceover for a demo, captions for a clip, or a talking-head explainer can do the whole job in a tab without signing up or paying, and without any of it touching a server. Given that the comparable hosted tools charge monthly and meter generations, a free local-first editor that covers voiceover, captions, avatars, and export is worth an hour of anyone's evaluation, provided they go in knowing it's free rather than open source and that the first run of each model takes a moment.