Say when a take has been cleared instead of failing the download

Make polish real post-processing; fix distortion, underline and dead MP3 click

Stream takes from the server instead of buffering them in the page

Fix cursor flicker from a resizing modal; retire the bundled examples

Add playback polish; stop warm-up polluting generation time

Interpolate progress inside the current wave

Cache-bust static assets with content hashes

Put generation progress front and centre, and lock the workspace

Real generation progress: wave bar, ETA, elapsed, tab title; taller log

Stride the waveform peak scan so hours-long takes stay instant

Offer an MP3 download alongside the WAV

Replace the canvas waveform with wavesurfer.js + streaming RMS peaks

Style the import dialog's file picker

Add a 60s stall watchdog to the take download

Never orphan a finished take: server keeps it, UI offers recovery

Stream the finished-audio download with MB progress

Raise Modal generation timeout to 2 hours

Add a Stop generation button

Never fail a completed take on browser-side waveform decode

Give the Chorus wordmark a logomark and more presence

Lower batching gate to 250 words; split monologues into even atomic pieces

Clone dialog: multi-select Use as, longer read scripts, live recording pulse

Simplify the clone dialog: name first, no Use as pills

Add optional read-along scripts to the clone recorder

Sentence-split oversized turns so solo monologues can batch

Add an Optimize audio option to the clone dialog

Make cloned voices a persistent, named personal voice library

Rework voice picking and cloning around the slot, not the grid

Icon affordances for the player: pop-out on the dock, x to close the stage

Hide the collapsed-composer strip when expanded

Collapse the composer after commit; add a Now Playing stage

Fix stale cache hits: order-sensitive speaker key + cache version bump

Tolerate an older Modal backend when no voice clone is in use

Replace the empty-prompt alert with an inline, context-aware hint

Redesign the studio UI from the voice-studio prototype

Wire per-speaker voice-clone uploads through the generate endpoint

Redesign the composer as a single input shell, not loose siblings

Modal backend: accept a per-speaker uploaded voice clone

Add dropdown chevrons, grow the scenario field, harden voice preview

Add per-IP rate limits and a generation concurrency cap

Add a target-length control for AI script generation

Surface HF billing failures as a clear message, not a raw error dump

Give input fields real contrast, not a subtle tint

Fix invisible scenario input field

Redesign as an app shell: Chorus voice studio, cream/ink palette

Rebuild frontend as FastAPI + hand-built UI, replacing Gradio

Remove stale footer note

Fix stale anchor link to demo section

Update README comments

Restore HF Space YAML frontmatter