Chat threads • Solo and Ensemble conversations are now saved as threads. Pick any one up where you left off from the new sidebar. • Pin, rename, and delete threads. Each thread gets a short auto-written summary. • Start a new chat or a new cast without clearing your history. Reasoning models • Models that think before they answer now work in Solo and Ensemble. Their internal reasoning stays out of the transcript and is never read aloud. • Turn Thinking on or off mid-episode. The change applies on the next turn. Ensemble • Skip your turn when the cast invites you in, instead of waiting out the countdown. • Copy, edit, or delete any line in the transcript. Edits steer where the conversation goes next. • Markdown tables now render properly in both Solo and Ensemble. Settings • Settings and Speaker Isolation now fit on smaller screens. • Load and Eject buttons for your local model, plus an option to load it automatically at launch. • Done closes Settings right away. It no longer waits on a model load. • The app now tells you when a newer version is on the App Store. Fixes • Fixed a number of bugs

Fixes and improvements • Long narrations now finish cleanly. Longer passages could previously garble near the end, cut off early, or on some Macs quit unexpectedly. • More reliable voice imports. Some custom voices were being saved without actually capturing the speaker, producing flat, halting audio that never finished properly. Imports now verify the voice was captured, and any affected voices you saved earlier will prompt you to re-import them. • Clearer voice recordings. Removed processing that was quietly distorting louder speech while recording. What you record is now what gets used, and the preview reflects it accurately. • Numbers with commas are read correctly. "45,607" is now spoken as forty-five thousand six hundred and seven rather than as two separate numbers, and prices like "$1,500" are fixed the same way. • Trademark, copyright, section, and similar symbols are now spoken aloud instead of being skipped. • Questions sound like questions. Scripts exported from Chat were losing question marks and commas, which flattened the delivery.

--Accidentally attached an already released build in the 1.5.9 released. The correct new build reflecting the updates below has now been correctly attached-- This update makes Chat and Ensemble feel more like a real production booth: • Attach images in Solo Chat when your model supports vision, with clear feedback if something didn’t go through • See what your local model can do (vision, tools, thinking) and control how much it “thinks” • Tune temperature and related generation settings per saved Solo system prompt • Cleaner chat bubbles, Markdown, mute, and smoother composer behavior • Introduced Ensemble’s "Director’s Chair" — change pace and turn order mid-scene, boot a speaker out, or give one character a private steer • Save and share casts as a file, then re-import them later (voices remap safely if a custom one is missing) • Join the conversation yourself when you set a character name and turn “include me” on • Compact frees room for long Ensemble scenes without wiping your transcript • Clearer local-model connection status, browse all downloaded models, and load the one you want from App Settings • Speech and exports skip emoji so voices stay clean (they can still show in the chat) Enjoying Mimika? A quick rating goes a long way. Thanks for your support!

This update makes Chat and Ensemble feel more like a real production booth: • Attach images in Solo Chat when your model supports vision, with clear feedback if something didn’t go through • See what your local model can do (vision, tools, thinking) and control how much it “thinks” • Tune temperature and related generation settings per saved Solo system prompt • Cleaner chat bubbles, Markdown, mute, and smoother composer behavior • Introduced Ensemble’s "Director’s Chair" — change pace and turn order mid-scene, boot a speaker out, or give one character a private steer • Save and share casts as a file, then re-import them later (voices remap safely if a custom one is missing) • Join the conversation yourself when you set a character name and turn “include me” on • Compact frees room for long Ensemble scenes without wiping your transcript • Clearer local-model connection status, browse all downloaded models, and load the one you want from App Settings • Speech and exports skip emoji so voices stay clean (they can still show in the chat) Enjoying Mimika? A quick rating goes a long way. Thanks for your support!

This update is all performance and reliability improvements under the hood: • Faster, smoother switching between tabs, even with long scripts • Your voice picks and speaker names now stay exactly as you set them when moving scripts between features • Cleaner, more consistent text formatting in the script editor • Fixed visual glitches around pop-up windows and sheets • Lots of small fixes and refinements throughout Enjoying Mimika? A quick rating goes a long way. Thanks for your support!

Create a custom voice from any video Ever heard a voice in a video and wished you could use it? Now you can. Drop a video right into the Voice Manager and Mimika finds every person speaking in it. Listen to each one, pick the voice you want, give it a name — done. No editing software, no audio skills needed. Mimika even strips out the music and background noise, trims the silences, and stitches the voice together for you. Adding voices is smoother all around Add voices as fast as you can drop them — every single one finishes now, even if you queue up ten in a row. If a voice ever ended up half-finished in the past, just open the Voice Manager and it repairs itself. And the new Recover from Disk section finds voice recordings you thought were lost — press play to hear each one, bring it back with a fresh name, or delete it for good. Hear exactly what you're saving The voice enhancement screen now plays it straight with you. Play B plays your voice at the exact loudness it will actually speak with, so there are no surprises after you save. Turning a quiet voice up? A new meter under the loudness slider tells you the moment you're pushing too hard — and even when you do, the sound gets gently smoothed instead of crackling. Reproducible voices with seeds You didn't ask for it, but Mimika now supports reproducible generations. A "seed" is the value behind a generation — it's why the same sentence can sound a little different each time. Loved a take? Assign that seed to your voice, and it'll speak the same way every time — in Single Voice, Multi-Talk, and AI Chat. Want the variety back? Remove the seed and every take is fresh again. Plus plenty of small fixes to make adding and enhancing voices feel effortless.

Speaker Isolator accuracy + re-voice quality overhaul. - Speaker detection you can actually tune — the Speaker Sensitivity slider's dead zone is gone (the full travel now does real work), a new "Re-detect speakers" button re-runs detection on the cached audio without repeating the slow separation step, phantom duplicate speakers are merged automatically, and "Number of Speakers" now genuinely merges the detected speakers down to the count you set. - Re-voiced speech tracks the original lips — systematic lip-sync drift eliminated (measured ~2.3 s/min → ~0.1 s/min on a 3-speaker test clip, no word more than ~0.7 s off): word-aware segment caps, diarization end-padding recaptures trailing words, and an internal timing-QA pass measures every re-voice and automatically re-renders drifting takes. - Cleaner re-voiced audio — the scratchy, robotic line starts under "Match original speaking pace" are fixed (onset-protected compression + improved WSOLA alignment); long sentences share their timing slack instead of clipping words at chunk boundaries; over-long synthesis takes are automatically re-rolled; segment ends are pulled back toward the original timing; and years/numbers are never split across synthesis boundaries

New: Record in Voice Manager - Capture a reference clip with your microphone and save it as a custom voice, right inside the app. New: Ensemble - a Chat sub-mode where you and multiple AI personas hold one shared, autonomous, voiced conversation. Cast written by a local LLM or by API; turn-taking via Director (AI-picked), round-robin, or weighted-random with first-name mention addressing; per-speaker sampling presets shown as live per-turn badges; an agreement-collapse mechanism to break a stale consensus; rolling-summary context window; mic barge-in; export to Multi-Talk / History / Markdown. New: Menu Bar & Read Aloud (optional, opt-in) - a menu-bar voice picker plus a system-wide "Read Selection Aloud" macOS Service. Big fixes