Changelog

Voice Design (Beta)

You can now design a custom voice for your scene partner by describing it in plain text. Tell us what you want — "A warm, middle-aged woman with a slight Southern accent" — and we'll generate voice options for you to preview and choose from.

• Describe the voice you want in your own words
• Preview multiple generated options before selecting
• Your designed voices are saved and available for reuse across characters

Voice Library

Browse and search from thousands of voices in the new Voice Library tab. Find the perfect match for any scene partner and preview any voice instantly before selecting it.

• Search thousands of voices by name or keyword
• Preview voices directly in the picker
• Load more results as you scroll

My Voices

Any custom or library voice you've used is automatically saved under My Voices at the top of the Presets tab. No need to search for the same voice twice — your favorites are always one click away.

Other Improvements

• Voice preview on the setup screen now always plays the correct selected voice
• If a voice you previously selected is no longer available, you'll see a clear message to choose a new one

Slate Recording

Introducing slate recording — a brand new way to record your introduction separately from your scene. No more re-recording "Hi, I'm Jane Smith reading for Pat" every time you want another take at the scene. Record your slate once, save it, and pair it with any scene performance.

• Record your slate independently from your scene — save time on every re-take
• Full camera preview with mirror toggle before you start
• Review your slate and re-record if needed before saving
• All your slates live in their own dedicated tab in the workspace

Compose Tape

New Compose Tape feature lets you combine any slate with any scene recording into a single final self-tape — ready to download and submit. Pick your best slate, pick your best scene, and we'll put them together for you automatically.

• Pick any slate + any scene from your project to combine
• Preview recordings inline before selecting — hover over a thumbnail to watch it
• Your composed tape is automatically processed and ready for playback and download
• Works across all browsers and devices

Take Numbers

Every recording now shows its take number — a concept every actor already knows. No more guessing which recording was the good one based on timestamps.

• Takes are numbered per type: Slate Take 1, Scene Take 1, Composed Take 1
• Numbers accumulate across the project — consistent and predictable
• Visible everywhere: sidebar, gallery view, compose dialog, and playback

Mirror Toggle

New flip button on the camera preview lets you toggle between mirrored and un-mirrored view. Mirrored by default (what you expect from a selfie camera), but you can switch to see yourself as the casting director will — useful for checking eyeline direction and blocking.

• Available in both scene and slate recording flows
• Persists through countdown and recording
• Only affects the preview — your recorded video is always un-mirrored

Other Improvements

• Project title now appears in gallery and compose dialog headers
• Recording cards show type, take number, and date on separate lines for a cleaner sidebar layout
• Light mode theming improved across review screens

Vision-Based Speech Detection

Smart Detect now uses real-time vision tracking as a third detection signal alongside speech-to-text and voice activity detection. When you finish speaking a line, the system can see that your mouth has closed — adding a layer of confirmation that makes line advancement significantly more accurate.

• Works automatically in Smart Detect mode — no setup required
• Analyzes your webcam feed in real time during recording
• Adapts automatically based on line length — shorter lines stay snappy, longer lines get more patience
• Requires sufficient speech coverage before it can trigger, preventing false advances during mid-line pauses

Long-Line Protection

Monologues and long speeches now have a dedicated protection system. On long lines, the system waits for multiple signals to confirm you've actually finished speaking before advancing. This means you can deliver a lengthy monologue without the system cutting off your last few words. Short and medium lines are unaffected — they still advance with the same snappy timing.

Recording End Buffer

When recording a self-tape, the system now holds the recording for a few extra seconds after your last line resolves. This ensures:

• Your final words are always captured, even if the system resolves slightly before you finish
• You get a natural "hold the beat" moment at the end — standard practice for professional self-tapes
• You can always hit Stop Recording to end sooner if you're done

Accuracy Improvements

• Speech matching thresholds have been retuned across all line lengths for better accuracy on long speeches
• Overall detection accuracy improved to 99% across tested scripts

Other Improvements

• Character names in the sidebar now show a tooltip on hover when truncated
• Notes card text overflow handling improved for long notes