模型需在设置中或通过 CLI 单独下载后才能使用；OmniVoice、YuE2、Breeze TTS 2 权重仅限非商业用途，Stable Audio 3 受 Stability AI 社区许可约束；开发模式需先按开发指南准备依赖。
yovoice
leemyswApache-2.0活跃维护本地离线的 AI 音频与音乐工作站，支持语音合成、声音克隆、带歌词歌曲和多轨乐谱生成，无需联网和 API 费用。
Open-source local AI audio & music studio for macOS & Windows. TTS, voice cloning, songs with lyrics, and multi-track scores — 100% offline, zero API fees.
v0.6.0发布于 2026-11-09 08:21 UTC2,158star +53 / 7 天161fork同步于 2026-11-10 23:46 UTC
make install
make app-run 硬件要求按 README 说明
README 未给出具体显存数值，仅说明可完全在本地 GPU 或 CPU 上运行；macOS 需 14+ 且为 Apple Silicon，Windows 为 10/11 x64，另有 Linux x64 版本，并随包提供 Windows CUDA 12.4/13.3 运行时与 macOS Metal 支持。
纯 CPU 可运行支持 Apple Silicon需要 CUDA
本地离线运行的语音与音乐创作工作室，支持 TTS、克隆音色、生成歌曲和多轨乐谱
本地离线的 AI 音频与音乐工作站，支持语音合成、声音克隆、带歌词歌曲和多轨乐谱生成，无需联网和 API 费用。
核心能力
- 零样本克隆参考音色并调节情绪与语气
- 按角色逐行分配台词生成故事配音
- 用 ACE-Step、YuE2、Stable Audio 3 生成歌曲与音效
- 按小节拍号编写乐谱并导出 MIDI 与分轨
- 内置 MCP 服务供 Cursor、Claude Code 调用
容易踩的坑
内容由模型根据 README 生成，可能不完整或不准确，以官方文档为准。看原文 README
- v0.6.0最新正式版2026-11-09 · 2026-11-09 08:21 UTC
[0.6.0] - 2026-11-09
English
- Speech recognition and subtitles with the optional Qwen3 ASR 0.6B model (30 languages): a new Transcribe page turns audio or video into timed subtitles (SRT / WebVTT); “Recognize text” fills in a reference transcript; speech versions can export aligned subtitles.
- Stems: “Separate stems” splits a music version into drums, bass, vocals, guitar, piano and other (HTDemucs, with optional BS-RoFormer for cleaner vocals). The new full-screen stem studio plays all stems in sync with per-track waveform, volume, mute and solo. Stems can be exported to a folder with DAW-friendly names, separated again or deleted, and songs up to 10 minutes now separate reliably.
- “Transcribe to MIDI” (optional MuScriptor model, non-commercial) turns a music version into a score project with one track per instrument and sensible levels, so the lead melody is no longer buried.
- YuE2: hum a melody into a score, edit or hand-write the melody score in a large editor, re-sing only the changed bars, make up to 4 candidate takes, apply LoRAs, take style tags from an image, and tune Top P / Top K / repetition penalty.
- Export audio as FLAC or M4A as well as WAV, with title, lyrics, cover and an AI-generated note in the tags; version exports can also target audiobook, podcast or video loudness.
- “Listen while generating” for models that support streaming; Settings › Engine › Runtime sets threads, GPU device, idle unload and model-specific options. Kokoro gets a speaking-rate control.
- Long reference recordings open a section picker with the clearest part preselected.
- Import existing local GGUF models and SF2 sound fonts without copying them; models replaced by a catalog update can be rolled back to the earlier version.
- Interface: smoother transitions, a resizable side panel, Transcribe moved under Text to speech, simpler Settings copy, less CPU when idle, and a smaller install with a faster first screen.
- Reliability: automatic restore from backup, safer media deletion and data-folder moves, resumable downloads that survive network hiccups, no hang on quit, errors shown where they happen (Ocho panel, music player, dialogs), and a clear message instead of a blank window when the macOS window fails to load.
简体中文
- 可选语音识别模型 Qwen3 ASR 0.6B（支持 30 种语言）：新增“语音转写”页，把音频或视频转成带时间的字幕（SRT / WebVTT）；“识别文字”自动填写参考原文；语音版本可导出对齐的字幕。
- 分轨：“分离音轨”把音乐版本分成鼓、贝斯、人声、吉他、钢琴和其他六轨（HTDemucs，可选 BS-RoFormer 让人声更干净）；新的全屏分轨工作台统一播放各轨，每轨有波形、音量、静音和独奏。分轨可按 DAW 习惯命名导出到文件夹，也可重新分离或删除；10 分钟以内的歌都能稳定分离。
- “转为 MIDI”（可选 MuScriptor 模型，仅限非商业用途）：把音乐版本转成编曲作品，每种乐器一轨并按角色定电平，主旋律不再被伴奏盖住。
- YuE2：哼一段扒成旋律谱，旋律谱可放大编辑或手写，按改动只重唱变化的小节，一次最多生成 4 个候选，支持 LoRA，可看图取风格，并可调写谱阶段的 Top P / Top K / 重复惩罚。
- 导出音频新增 FLAC 和 M4A，标签里写入标题、歌词、封面并注明 AI 生成；版本导出也可选有声书、播客或视频平台的目标响度。
- 支持流式的模型可“边生成边试听”；“设置 › 推理引擎 › 运行参数”可设置线程、GPU 设备、空闲卸载和模型专属参数；Kokoro 新增语速调节。
- 较长的参考录音先选段，已预先选中最清楚的一段。
- 可原地导入本机已有的 GGUF 模型和 SF2 音色库；经目录更新替换过的模型可以回退到旧版本。
- 界面：过渡更连贯，右侧面板可调宽，语音转写移到语音生成下方，设置页文案更简洁，空闲更省 CPU，安装包更小、首屏更快。
- 更可靠：状态库丢失时自动从备份恢复，删除素材和迁移数据目录更稳妥，下载遇网络抖动自动续传，退出不再卡住，错误在出错的地方就地提示（Ocho 面板、音乐播放页、各对话框），macOS 窗口加载失败时给出原因而不是白屏。
贡献者 / Contributors
@leemysw @zoyluoblue @david-bowiegxw
- v0.5.02026-10-28 · 2026-10-28 10:14 UTC
查看更新说明
[0.5.0] - 2026-10-28
English
- 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 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).
- 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.
- Kokoro can blend two voices of the same language, set per story-narration role.
- 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 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.
- Export to a target loudness: audiobook −18, podcast −16 or video platforms −14 LUFS, with peaks kept under −1 dBTP.
- 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 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".
- Smoother background-music ducking: it dips before speech starts, holds through short pauses and recovers smoothly.
- The Settings › Resources table is simpler and fits on one screen, with row actions in a "More" menu; narrow windows scroll sideways.
- 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: the asset dialog in story narration keeps a stable height when switching types, and the editor scrollbar sits against the divider.
简体中文
- 新增 6 个语音模型：Ming 3.0 TTS、Voxtral TTS、Plapre 3、XTTS v2、Gradium Phonon 和 Fish Audio Drama 3，覆盖内置音色、声音设计与多语言音色克隆，Drama 3 还支持说话人和事件标签；故事配音的每个角色可分别使用，需要 audio.cpp v0.9.3 内核。
- 音乐生成支持 Stable Audio 3（Small Music、Small SFX、Medium，Q8 / F16）：按英文描述生成纯音乐，Small 最长 2 分钟、Medium 约 6 分钟，可以已生成的版本为起点整体改编或局部重绘；新增“音效生成”，按描述生成 1–30 秒音效。模型采用 Stability AI 社区许可，年收入低于 100 万美元可免费使用。
- 推理内核升级到 audio.cpp v0.9.3：修复 Metal 与 CUDA 上偶发的计算错误，多个模型生成更快；已装 v0.7.4 及以上内核的仍可使用，设置页可选更新。CUDA 新增“性能优先”模式，为 ACE-Step、Qwen3-TTS、YuE2、VoxCPM2 和 Drama 3 加速，但占用更多显存。
- Kokoro 支持按比例混合两种同语言音色，故事配音的每个角色可分别设置。
- 已下载的模型在上游发布新版本后，设置里标出“有新版本”并可一键更新，侧栏给出一次可关闭的提示；旧版本仍可使用。
- 生成后自动体检音频：发现爆音、咔哒声、结尾截断、长静音、几乎无声或音量过小时，在播放栏和故事配音对应句子旁提醒，悬停可看问题和时间位置；Ocho 交付前也会检查，故事配音只重新生成有问题的句子。
- 导出时可选目标响度：有声书 −18、播客 −16、视频平台 −14 LUFS，峰值不超过 −1 dBTP。
- 参考音频支持最长 10 分钟的录音：超过 60 秒时自动截取人声最清楚的 5–15 秒，裁剪时也可点“自动挑选”。
- 故事配音可直接把 yovoice 生成的语音、音效和音乐加到音轨；Ocho 能给故事配音加音效、更新模型和开启“性能优先”。
- 背景音乐压低更自然：说话前提前让位，句间短停顿不再忽高忽低，说完后平滑恢复。
- 设置“资源”页表格精简，一屏放下，每行操作收进“更多”菜单；窄窗口可左右滚动。
- 提示更清楚：Ocho 面板选中 ChatGPT 订阅时直接说明助手用不了并可就地切回；推理引擎中途崩溃时提示“推理引擎意外退出”并附上原因，不再显示像网络故障的报错。
- 界面细节：故事配音素材弹窗切换类型时高度保持稳定，作品编辑区滚动条贴靠右侧分割线。
- v0.4.12026-10-21 · 2026-10-21 10:41 UTC
查看更新说明
[0.4.1] - 2026-10-21
English
- 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.
简体中文
- 修复 macOS 自动更新下载完成后总是提示“更新校验失败（codesign）……invalid requirement specification”、无法安装的问题。问题出在 0.3.0–0.4.0 自带的更新程序里，macOS 用户请手动下载安装一次 0.4.1，之后即可正常自动更新。
- v0.4.02026-10-21 · 2026-10-21 09:19 UTC
查看更新说明
[0.4.0] - 2026-10-21
English
- New Linux x64 desktop app:
.deb(Debian/Ubuntu) and.tar.gzwithinstall.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. - 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.
- 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.
- 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.
- Smaller installers: the Kokoro pronunciation data (about 10 MB) now downloads together with the Kokoro model.
- 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.
- 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.
- 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".
- Fix model downloads from ModelScope failing size checks after the mirror was refreshed.
- The Windows installer now supports Windows 10 1809 (17763) and later.
- On upgrade, saved projects, assistant chats and engine records are converted to the new format once, with a database backup made first.
简体中文
- 新增 Linux x64 桌面版：提供
.deb（Debian/Ubuntu）和附install.sh的.tar.gz，开箱使用内置 CPU 内核，可在设置中下载 Vulkan GPU 内核；需要系统提供 WebKitGTK 4.1，暂不支持自动更新。 - 音乐播放页逐字点亮歌词：按音频里的人声分出每个字的时间；新增一句一屏的“浮光”样式，字随演唱逐个飞入亮起，适合录屏；可导出逐字 LRC。
- 新增可选下载的歌词对齐模型（BS-RoFormer 人声分离 + Qwen3 Forced Aligner，约 1.3 GB）：“智能对齐”先分离人声，去掉间奏和长停顿，再按句把歌词逐字对到演唱上，结果随版本保存。
- 音乐播放页焕新：macOS 上标题栏与背景融为一体；校准、对齐、导出收进“歌词工具”菜单；“浮光”动效和背景重做，歌词滚动更顺滑。新增“调整歌词时间”：把一句锚定到当前播放位置，或整句提前、推后 0.1/0.5 秒（←/→），可选后面跟着移或只改这一句；单字可在时间轴上拖动，也可边听边跟唱打点。
- 安装包变小：Kokoro 发音库（约 10 MB）改为第一次下载 Kokoro 模型时一并下载。
- 设置 › 通用新增“功能显示”，可分别隐藏故事配音、音乐生成和 Ocho 助手；Ocho 与音乐生成都关闭时一并隐藏 AI 服务设置。
- 启动更快、界面更流畅：修复最近作品里有二三十分钟长音频时启动停在启动画面的问题；各页面第一次打开时才加载；生成、下载和 Ocho 回复过程中不再重绘整个界面。
- 修复 Windows 和 Linux 上导出约两分钟以上的音乐时点击没有反应的问题，时间线导出长音频也不再受限；播放器工具栏改为“打开位置”，导出移到“更多”菜单。
- 修复 ModelScope 镜像刷新后，从该来源下载模型时大小校验失败的问题。
- Windows 安装包最低系统要求放宽到 Windows 10 1809（17763）。
- 升级时一次性把旧版本保存的作品、助手对话和内核记录转换为新格式，转换前自动备份数据库。
- New Linux x64 desktop app:
- v0.3.12026-10-20 · 2026-10-20 01:14 UTC
查看更新说明
[0.3.1] - 2026-10-19
English
- Fix a crash on launch on macOS 26 caused by the sidebar button in the title bar.
- Fix long pages on Windows being covered by a large black block; main scroll areas now show a thin, light scrollbar.
简体中文
- 修复 macOS 26 上标题栏侧边栏按钮导致应用启动后闪退的问题。
- 修复 Windows 上内容超过一屏的页面被一大片黑色遮住的问题；主要滚动区域显示细的浅色滚动条。
用户点评
登录用户的真实使用记录，可按版本和硬件筛选共 0 条
还没有人点评。第一个说出真实体验的人最有价值。
yovoice is an open-source, local-first audio and music creation studio for macOS, Windows and Linux.
Turn text into expressive speech, natural cloned voices, complete vocal songs, and multi-track scores right on your machine. Featuring zero-shot voice cloning, emotional performance controls, AI music generation with lyrics, and General MIDI score arranging—all running 100% locally with zero cloud API calls, zero per-character fees, and zero Python environment headaches.
Why yovoice?
- 🔒 100% Local Privacy: Runs entirely on your local GPU/CPU. Your manuscripts, reference voices, and creations never leave your computer.
- 💸 Zero Recurring Fees: Say goodbye to per-character/per-minute meters and expensive cloud voice/music subscriptions.
- ⚡ Native C++ Performance (No Python Hell): Powered by audio.cpp v0.9.3. Ships with bundled Windows CUDA 12.4/13.3 runtimes and native macOS Metal support. Double-click and create—no Conda or PyTorch conflicts.
- 🎛️ Creator Workstation: Multi-take versioning for side-by-side comparison, waveform trimming, timeline scrolling, and clean stems/MIDI export.
- 🤖 MCP & AI Agent Ecosystem: Chat with DeepSeek, Qwen, Claude, OpenAI, or local Ollama to create and edit works end to end. Includes a native Model Context Protocol (MCP) server for Cursor and Claude Code.
Features
- Speech, voice design and cloning — clone a reference voice, design a voice from a description, adjust emotion and delivery, and edit takes on a timeline.
- Story dubbing — import text or subtitle files, assign a character to each line, and generate line by line.
- Music — write a style and lyrics with section markers, then generate songs or instrumentals with ACE-Step 1.5, or songs and score-based instrumentals with YuE2, or instrumental music and sound effects from an English description with Stable Audio 3; the playback page syncs lyrics to the vocals.
- Score — write parts by bar and beat, import MIDI or let an Agent compose, render with a General MIDI sound font, and export MIDI and stems. See Score.
- Optional AI service — the built-in assistant panel chats with DeepSeek, Qwen, Kimi, OpenAI, Anthropic, Ollama or any compatible API and acts on your works: it writes arrangements and lyrics, edits text, assigns voices and starts generation; keys stay on this computer. See AI service.
- 12 interface languages — English, 简体中文, 繁體中文, 日本語, 한국어, Español, Français, Deutsch, Português (Brasil), Русский, Italiano and Tiếng Việt; switch anytime in Settings, no restart needed.
Supported Models
| Use | Models |
|---|---|
| Voice cloning and emotion | IndexTTS 2.0 / 2.5, Qwen3-TTS Base |
| Voice design and cloning | VoxCPM2, OmniVoice, Qwen3-TTS VoiceDesign, Breeze TTS 2 |
| Built-in voices | Kokoro-82M, Qwen3-TTS CustomVoice |
| Songs and instrumentals | ACE-Step 1.5 Turbo / XL Turbo, YuE2 3B |
| Instrumentals and sound effects | Stable Audio 3 Small Music / Small SFX / Medium |
All models are downloaded inside the App or with the CLI. See model selection for file sizes, hardware requirements, inputs and examples. OmniVoice, YuE2 and Breeze TTS 2 weights are restricted to non-commercial use; Stable Audio 3 is free below USD 1M annual revenue under the Stability AI Community License.
Installation
Choose the package for your platform from the repository’s Releases tab or visit the official website.
| Platform | Package | Install |
|---|---|---|
| macOS 14+ · Apple Silicon | .dmg | Open the disk image and drag yovoice to Applications |
| Windows 10/11 · x64 | -setup.exe | Run the installer, then open yovoice from the Start menu |
| Windows 10/11 · x64 · NVIDIA GPU | -cuda12.4-setup.exe | Bundles the CUDA 12.4 runtime; works with older GPUs and drivers and uses the GPU from first launch |
| Windows 10/11 · x64 · NVIDIA GPU | -cuda13.3-setup.exe | Bundles the CUDA 13.3 runtime; needs a newer driver, smaller download |
| Linux · x64 (Debian/Ubuntu) | -linux-x64.deb | sudo apt install ./yovoice-<version>-linux-x64.deb, then open yovoice from the application menu |
| Linux · x64 (other distributions) | -linux-x64.tar.gz | Extract and run ./install.sh to add yovoice to the application menu for the current user |
The Windows installer downloads and installs WebView2 Runtime if needed. Models are downloaded inside the app; uninstalling preserves user data in ~/.yovoice.
On Linux the app needs WebKitGTK 4.1 (libwebkit2gtk-4.1-0 on Debian/Ubuntu, webkit2gtk4.1 on Fedora, webkit2gtk-4.1 on Arch); the .deb installs it automatically. Linux runs on the CPU out of the box, and a Vulkan GPU runtime can be downloaded in Settings → Inference engine. The Linux app doesn't update itself yet; download new versions from Releases.
Check for updates from the app menu on macOS or Windows. Updates are also checked and downloaded in the background; restart to install when ready. Windows updates download only the app (-update.exe) and keep the installed runtime; when a release recommends a newer runtime you can upgrade it in Settings → Inference engine.
Quick Start
- Set up a model. Download a model in Settings (e.g. Kokoro-82M for fast narration, or VoxCPM2 for cloning).
- Create a project. Choose Story dubbing, Speech generation, Music, or Score.
- Add content. Enter text, lyrics, or notes, then pick a character and adjust style.
- Generate audio. Select Generate, then preview, compare takes, edit, and export.
CLI & Agent Skill
Give your agent the yovoice Skill link and ask it to install the Skill and set up the local CLI, engine, and model:
Install this Skill and set up yovoice for local speech generation: https://github.com/leemysw/yovoice/tree/main/skills/yovoice
Then describe what you want:
Read narration.txt using voice.wav as the reference voice, with a calm delivery, and save it as narration.wav.
The agent runs the standalone CLI without opening the desktop app. See the CLI guide for details.
Run yovoice serve to provide an authenticated HTTP API and MCP inference service for other machines. See the API / MCP deployment guide for deployment and client examples.
Development
make install
make app-run
See the development guide for prerequisites. It also covers browser preview, tests, and project structure. For how the processes and packages fit together, see Architecture.
Community
Scan the QR code with WeChat to join the yovoice user group. If the code has expired, open an issue and we will post a new one.
Contributing
Bug reports, feature suggestions, and pull requests are welcome. Include your platform, model, and steps to reproduce when reporting a problem. Run make check before submitting code changes.
Acknowledgements
- audio.cpp by ShugoAI — the local audio inference engine.
- VoxCPM — voice design and cloning models, licensed under Apache-2.0.
- IndexTTS — the speech synthesis models behind yovoice.
- OmniVoice — voice design and cloning models.
- Qwen3-TTS — voice cloning, built-in voices, and text-guided voice design.
- ACE-Step — music generation from style prompts and lyrics.
- YuE — full vocal song generation from lyrics.
- Stable Audio 3 by Stability AI — instrumental music and sound effects from text. Powered by Stability AI.
- Breeze TTS 2 — Chinese and English voice design and cloning.
- go-meltysynth, MuseScore General, GeneralUser GS and SGM-V2.01 by David Shan — the SoundFont synthesizer and sound fonts behind score rendering.
- Kokoro — lightweight speech synthesis with built-in voices: Kokoro-82M and Kokoro-82M-v1.1-zh.
Star History
License
Apache-2.0. Dependencies retain their original licenses; IndexTTS models are covered separately by their model license.
whisper
openai·Python
OpenAI 开源的通用语音识别模型，基于大规模弱监督训练，支持多语言转写、翻译和语种识别，开箱即用。
语音识别+4语音进阶
112371+23414kv202506252026-08-31 17:19 UTC
GPT-SoVITS
RVC-Boss·Python
只需 1 分钟语音数据就能训练出高质量 TTS 模型，支持少样本音色克隆，适合快速定制个性化语音合成。
语音合成+4语音硬核
64721+2816.8k20250606v2pro2026-11-09 13:59 UTC
Real-Time-Voice-Cloning
CorentinJ·Python
用 5 秒语音样本克隆音色，实时合成任意文本的语音，基于 PyTorch/TensorFlow 的说话人编码与声码器流水线。
语音克隆+4语音进阶
60236+69.4k2026-03-09 10:31 UTC
VoxCPM
OpenBMB·Python
无需分词器的多语言语音合成模型，支持创意音色设计和逼真声音克隆，可直接从文本生成自然语音。
语音合成+4语音进阶
42209+10654.8k2.0.32026-11-09 07:36 UTC