tautau.
Back to docs

Edit video and audio in Studio

Preview

Drop a video and a separate audio recording into the editor. Studio loads editable tracks, checks for matching sound and gives you a shared timeline to make the cut. Keep your original files: the project lives in this browser.

Open Studio

Make your first edit

  1. Open Studio and sign in if asked. Drop local video and audio over the editor, or choose Import media. You can choose several files together.
  2. Watch the loading status and open File results. Each valid source gets its own track; a failed file does not prevent the other valid files from loading.
  3. Leave Sync matching audio checked to try local alignment. Read the result and play the tracks together. Short, silent or uncertain sources stay editable at their imported positions.
  4. Select a clip. Drag it along its lane to move it, or set Timeline start in the Inspector. Set Source in and Source out to trim, then apply the Inspector changes.
  5. Move the playhead inside the selected clip and choose Split. Use Undo and Redo while you review. Add captions or reviewed suggestions when the timing is right.
  6. Choose the whole composition or a selected range, pick an output format and export. Use Save project to download an edit backup, and keep the original media alongside it.

Load and align sources

Matching is on by default for local files. Turn it off before importing if you want to place everything yourself. Studio compares 12–60 seconds of source audio locally within a ±15-second search and moves only reliable, distinct matches. It does not upload audio for matching, correct clock drift, trim sources or choose which microphones to hear.

A batch uses an eligible audio file as its reference, falling back to audio in a video. If you add one file later, Studio tries an existing reference and keeps your earlier edit fixed. A match that would place the new clip before timeline zero needs manual alignment. The result names the reference and explains files it could not compare.

When matching changes positions, loading and alignment are separate undo steps: Undo alignment keeps the files and tracks. Cancel audio matching also keeps the loaded sources. If the files are already aligned, positions stay unchanged. For an uncertain result, move the new clip manually or use the reviewed camera workflow below.

Edit and mix tracks

Select, move, trim, split, duplicate or delete clips on the shared timeline. The media bin lets you reuse a source with Add to track. Rename tracks, add lanes and reorder video tracks; the top active video track supplies the picture. Edits keep the original source intact.

Enabled clip audio mixes together. Adjust clip, track and master gain, fades and low-cut filtering, and watch the monitor peak for overload. Matching does not mute duplicate microphones: mute unwanted tracks or turn off Include clip audio. Solo and repair bypass are temporary listening controls; export uses the saved mix.

For a local pop or rumble, select the clip and set a repair region in original source time. Compare the attenuated or filtered region with the current mix before keeping it. Repairs can be disabled, removed and undone. They are local edits, not general noise removal.

Review camera edits

Open Synchronise sources for a separate camera workflow: one master audio recording and two to four cameras. Mark the same event in each original source, or review suggested audio anchors. Preview the proposed positions before applying them.

Choose shots on the picture-only camera programme while the master audio continues. Camera inputs exclude their audio by default. Grouped sources move together; use the group controls for a source-preserving trim, or detach a source to edit it independently. This workflow does not choose shots automatically or correct drifting clocks.

Add captions and review AI suggestions

Select a clip and transcribe a source range. This sends the selected audio to the transcription service and its speech provider; transcription usage and supported speech-to-text API-key settings apply. Click timed words to seek. Caption corrections change the caption text while preserving the raw transcript. Export SRT or VTT subtitles mapped to the current timeline. If the provider supplies no word timings, Studio keeps the text untimed rather than inventing alignment.

Suggestions need your review

  • Cuts: find literal fillers locally, or ask AI for cut proposals from timed transcript words. Preview the proposed cut, then accept individually or as a batch. Accepted cuts affect the shared timeline and can be undone.
  • Audio repairs: bounded local analysis measures possible transients. AI receives measurements and the current-effects summary, not audio, and can recommend a fixed low-cut or attenuation preset. Preview the current and proposed mix and acknowledge each repair before applying it. Even a low-uncertainty choice needs listening and explicit acceptance. These measurements do not diagnose a particular speech sound or clipping in the original recording.
  • Highlights: AI uses provider-timed words to suggest 5–60-second excerpts. Preview or export an excerpt across the tracks with captions rebased to its start. This does not change your saved composition or undo history.

AI cuts and highlights send transcript words, a source label and your instruction to the configured model provider. Repair requests send audio measurements instead. Suggestions use a server-selected model; speech-to-text BYOK does not select the suggestion model. Nothing is applied automatically. A cancelled, stale or invalid response leaves your edit unchanged.

Bring in library media or a link

Import from tautau lists ready recordings and audio/video assets in your current personal or workspace library. Importing makes a verified local snapshot. Cloud changes do not update that copy, and your cloud original stays unchanged.

Import URL accepts a directly downloadable media file, including a podcast enclosure, up to 50 MiB. It first creates a private cloud library asset using your asset storage allowance, then copies the file into this browser. Cloud storage and browser storage are separate. Feed pages, YouTube pages and adaptive playlists do not work in this action. Creating a cloud copy needs workspace write access.

The optional Import YouTube action uses a separately configured extractor. Choose an explicit original start and end for a 50 ms–120-second excerpt, capped at 50 MiB and within the original video's first 24 hours. Availability depends on the source and the configured service. The acquired video is a private cloud asset followed by a local snapshot. Its source card retains the requested original range, but editing starts at zero in the excerpt's own clock. A transcript alone is not editable video.

Stop waiting or Cancel local copy does not cancel or delete a cloud job. Use the matching Resume import action to check that request again, or find the ready item in your library. A stopped local copy may need to be imported again. Cloud imports can fail because of quota, library capacity, source availability or browser codec support; read the error before starting a new request.

Export the whole edit or a range

Choose Video · WebM for 720p picture and mixed audio, or Audio · WebM for sound only. Set a timeline range to export just that interval; silence and gaps remain in the output, and sidecar captions start at the range's local zero. WebM export runs in real time. Keep the tab visible until it finishes.

Audio · WAV produces 48 kHz stereo, 16-bit PCM for selections of 50 ms–60 seconds. It supports PCM WAV, AAC in MP4/M4A and Ogg Opus sources. MP3 and WebM/Matroska sources need WebM output in this preview. If the rendered WAV clips, Studio holds it for an explicit download or discard decision. Filters restart at a selected range's start, so its opening can differ from a crop of the full mix.

Preview limits
LimitCurrent allowance
Source files128 MiB each; 50 ms–2 hours of source duration; 512 MiB retained in a project
Timeline16 sources, 16 tracks, 32 clips, 8 active at once
Composition10 minutes; longer sources initially add their first 10 minutes
ExportsWebM up to 10 minutes; supported WAV mixes up to 60 seconds

Codec support depends on your browser. Desktop Chromium is the tested path; broad browser support and frame- or sample-exact editing are not promised. If a source cannot be decoded, try a supported copy while retaining your original.