A record of every step in yovoice's evolution. Built local-first and open source, we continuously expand the possibilities of speech synthesis, voice cloning, and music composition—giving your words a voice with genuine emotion.
Speech Models: Six new speech models: Ming 3.0 TTS, Voxtral TTS, Plapre 3, XTTS v2, Gradium Phonon and Fish Audio Drama 3, covering built-in voices, voice design and multilingual voice cloning; Drama 3 also takes speaker and event tags. Each story-narration role can use them separately. They need the audio.cpp v0.9.3 engine.
Music Generation: Music generation adds Stable Audio 3 (Small Music, Small SFX, Medium; Q8 / F16): generate instrumentals from an English description, up to 2 minutes with Small and about 6 with Medium, remix a whole existing version, or redraw part of it. New "Sound effects" generates 1–30 s effects from a description. Stable Audio 3 uses the Stability AI Community License (free below US$1M annual revenue).
Inference Engine: The engine is upgraded to audio.cpp v0.9.3: fixes occasional wrong results on Metal and CUDA and speeds up several models. Engines v0.7.4 and later keep working; update from Settings. A new CUDA "Performance first" mode speeds up ACE-Step, Qwen3-TTS, YuE2, VoxCPM2 and Drama 3 at the cost of more VRAM.
Voice Blending: Kokoro can blend two voices of the same language, set per story-narration role.
Model Updates: Downloaded models show "Update available" in Settings when upstream releases a new version, with one-click update and a dismissible sidebar notice; old versions keep working.
Audio Quality Check: Audio check after every generation: clipping, clicks, cut-off endings, long silences, near-silence or low volume are flagged in the player and next to the affected story lines, with details and timestamps on hover. Ocho checks audio before handing it over and regenerates only the faulty story lines.
Target Loudness: Export to a target loudness: audiobook −18, podcast −16 or video platforms −14 LUFS, with peaks kept under −1 dBTP.
Reference Audio: Reference recordings up to 10 minutes: clips over 60 s are trimmed to the clearest 5–15 s of speech automatically, and "Auto pick" is available when trimming.
Story Narration: Story narration can add yovoice-generated speech, sound effects and music straight to a track; Ocho can add sound effects to a story, update models and turn on "Performance first".
Audio Ducking: Smoother background-music ducking: it dips before speech starts, holds through short pauses and recovers smoothly.
Resource Management: The Settings › Resources table is simpler and fits on one screen, with row actions in a "More" menu; narrow windows scroll sideways.
Error Messages: Clearer errors: picking a ChatGPT subscription in the Ocho panel explains right away that the assistant can't use it and lets you switch back; an engine crash mid-generation now says "The inference engine exited unexpectedly" with the reason, instead of a network-like error.
Interface Polish: Interface polish: the asset dialog in story narration keeps a stable height when switching types, and the editor scrollbar sits against the divider.
v0.4.1 2026-10-21
Auto-Update: Fix macOS auto-update always failing after download with "Update verification failed (codesign) … invalid requirement specification". The bug is in the updater of 0.3.0–0.4.0 itself, so on macOS please download and install 0.4.1 manually once; later updates will install automatically.
v0.4.0 2026-10-21
Linux Platform: New Linux x64 desktop app: .deb (Debian/Ubuntu) and .tar.gz with install.sh. It runs on the bundled CPU engine, can download a Vulkan GPU engine in Settings, and needs WebKitGTK 4.1; auto-update is not available on Linux yet.
Synced Lyrics: Word-by-word lyrics on the music playback page: each word lights up as it is sung, timed from the vocals. A new one-line-per-screen "Glow" style flies words in and lights them up, good for screen recording. Export word-level LRC.
Smart Alignment: Optional lyric alignment model (BS-RoFormer vocal separation + Qwen3 Forced Aligner, about 1.3 GB): "Smart align" separates the vocals, skips interludes and long pauses, and aligns the lyrics word by word, line by line; results are saved with the version.
Playback Page: Refreshed playback page: an immersive title bar on macOS, a "Lyric tools" menu for calibration, alignment and export, reworked "Glow" motion and backdrop, and smoother lyric scrolling. "Adjust lyric timing" anchors a line to the playhead or nudges it by 0.1/0.5 s (←/→), with or without shifting later lines; drag single words on a timeline or tap along while listening.
Compact Packages: Smaller installers: the Kokoro pronunciation data (about 10 MB) now downloads together with the Kokoro model.
Feature Toggles: Settings › General › "Features" lets you hide story narration, music generation and the Ocho assistant; AI service settings hide when both Ocho and music are off.
Performance: Faster launch and a smoother interface: fixes launch stalling on the splash screen when recent work includes 20–30 minute audio; pages load on first open; generation, downloads and Ocho replies no longer redraw the whole interface.
Export Fix: Fix exporting longer music (about two minutes or more) doing nothing on Windows and Linux; long timeline exports are no longer limited either. The player toolbar now has "Show in folder", with export under "More".
Mirror Downloads: Fix model downloads from ModelScope failing size checks after the mirror was refreshed.
Compatibility: The Windows installer now supports Windows 10 1809 (17763) and later.
Data Migration: On upgrade, saved projects, assistant chats and engine records are converted to the new format once, with a database backup made first.
v0.3.1 2026-10-19
Stability Fix: Fix a crash on launch on macOS 26 caused by the sidebar button in the title bar.
Display Fix: Fix long pages on Windows being covered by a large black block; main scroll areas now show a thin, light scrollbar.
v0.3.0 2026-10-19
AI Assistant: Meet Ocho, the built-in AI assistant (⌘J / Ctrl+J): create and edit projects, write lyrics and scores, pick voices, tune parameters and start generations by chatting. It sees your selection and recent results, remembers preferences, and asks before anything destructive.
Melody & Scores: YuE2 songs can start from a melody score: write or extract the melody and chords, view them as staff notation, tracks or ABC, preview with a sound font (notes light up as they play), edit, undo, export .abc, then generate from the score. Adds instrumental songs, covers via SheetSage2, and song forms that vary from song to song.
Novels & Subtitles: Novel import for story narration (TXT, Markdown, EPUB), SRT/WebVTT subtitle export, LRC lyrics export, and side-by-side comparison of song versions on the playback page.
Resources & Logs: New settings pages: Resources to find and clean up unused audio, Logs with diagnostic bundles, separate Speech and Music model pages with one row per model and a precision switch, and ChatGPT (Codex) as an AI provider.
Multilingual Support: The interface now speaks Traditional Chinese, Japanese, Korean, Spanish, French, German, Portuguese (Brazil), Russian, Italian and Vietnamese, with an opening animation, a feature tour and smoother motion throughout.
Defaults & API Docs: New projects default to downloaded models; yovoice serve serves interactive API docs at /docs/; Score projects move under Music as "Score generation".
v0.2.1 2026-10-16
Direct Downloads: Add direct Kokoro-82M 1.1-zh downloads from Hugging Face and ModelScope, and simplify model management to downloads from the app or CLI.
Music Playback: Add a full music playback page with synchronized lyrics, vocal alignment and calibration, plus improved score limits, AI naming, and style editing.
SQLite Storage: Move projects, voices, models and settings to transactional SQLite storage with backups, migration from state.json, storage relocation, recovery, and lifecycle cleanup.
Desktop Reliability: Improve desktop reliability across Windows and macOS, including service recovery, process cleanup, tray behavior, narrow-window layouts, localized Windows UI, signing checks, and automatic updates.
Format Imports: Add Breeze TTS 2, text and subtitle file import, local MCP and API integration improvements, OrcaRouter, and clearer audio, migration, disk, and save-error handling.
v0.2.0 2026-10-11
Music Generation: New Music projects: set the style with tags, write lyrics with section markers or go instrumental, and generate full songs with ACE-Step 1.5 Turbo / XL Turbo or YuE2 3B (vocal songs whose length follows the lyrics; CC BY-NC 4.0, non-commercial). Each take is kept as a version you can preview, compare, reuse and export. Models made of several files can now be downloaded.
Score Projects: New Score projects: write notes per track by bar and beat, import MIDI, or let an agent compose, then render every track with a General MIDI sound font and mix by role. The track area spans the bottom of the window with sections on the timeline, zoom, scrolling and a playhead; export MIDI or score JSON. Choose MuseScore General, GeneralUser GS or SGM-V2.01. The CLI adds yovoice score render|midi|from-midi with stems, MCP adds render_score, and the Agent Skill adds a score guide.
AI Providers: Optional AI providers (DeepSeek, Qwen, Zhipu, Kimi, Doubao, MiniMax, OpenAI, Anthropic, OpenRouter, coding plans, local and custom endpoints) write a whole score or song style, lyrics and title from one sentence, run in the background, and can switch models on the spot.
Interface Polish: A refreshed interface: ticket-style project lists with record and cassette covers, grouped settings, a separate sound font page, side panel toggles in the title bar (the right panel toggle only appears in a workspace), and hover hints only for truncated text or icon buttons.
Inference Engine: The audio.cpp engine is upgraded to v0.9.1 and updates separately from the app; new Windows CUDA 12.4 and 13.3 installers.
Bug Fixes: Fixes for long AI requests timing out, saving on exit, Windows time-zone saves, timeline menus, deleting the last project, and clearer error messages and diagnostic logs.
v0.1.7 2026-10-05
Remote Service: Run a remote speech service with HTTP API and MCP support. Submit, query, cancel, and download generation tasks with authentication, retry deduplication, and configurable timeouts.
Agent Skill: Use the Agent Skill with either a local CLI or a remote service. User guides and CLI packages now include separate English and Chinese documentation.
Save Reliability: Fix Windows save-error reporting and allow a recovery backup when saving on exit fails. Improve diagnostic logs for saving and generation failures.
v0.1.6 2026-10-01
Performances: Save multiple performance styles per character, each with separate models and parameters, usable across story and speech projects.
UI Controls: Align and unify sizing for performance dropdowns, add, and more buttons.
Track Layout: Fix track resizing zone blocking play button; dragging triggers only near the top divider with more compact playback icons.
App Icon Polish: Remove highlight ring from app icon base, clean transparent edge artifacts, and calibrate macOS icon padding.
v0.1.5 2026-09-29
Story Projects: Introduce Story Narration and Speech Generation projects with subtitle import, segmented generation, and editable tracks.
Audio Editing: Add audio track trimming, undo/redo, mixing, and WAV export, with project archive export, import, and recovery.
Official Models: Add one-click download for official Kokoro BF16 models.
Release Notes: Add version changelog, automatically published to GitHub Release pages.
Playhead Tracking: Remove batch generation for pending lines; playback now continuously tracks current script lines.
Snapping & Scrolling: Improve track zooming, scrolling, clip dragging, and magnetic snapping; unify single and multi-role playheads.
Popover Polish: Fix dropdown popover styling on older WebKit; toasts no longer shift workbench layout.
Window Adaptation: Improve Windows window size and position persistence, plus macOS startup reliability and icon margins.
v0.1.4 2026-09-26
Character Library: Add Character Library to save and reuse voices and settings, sharing editing workflows with the studio.
Official Models: Add official Kokoro Q8 model download with 49 built-in voices; support experimental multilingual and Chinese imports.
Voice Management: Polish voice library, history, reference clips, and text transcripts with unified search, avatars, and modals.
Mic Permissions: Fix macOS microphone permission callback crash and improve compatibility with legacy dropdown selectors.
Encoder Fix: Fix VoxCPM2 encoder capacity limit when using longer reference audio clips.
Process Cleanup: Ensure background service waits for full cleanup before exit; retain Windows UI compatibility fixes.
v0.1.3 2026-09-19
Window Adaptation: Refine Windows window borders, scrolling, native menus, and dialogs.
