VocalFlow v0.8.0-beta.5

VocalFlow v0.8.0-beta.5

中文 | English

这是一个面向干净电脑安装测试的修复版。beta.4 的 VocalFlow Studio 安装包遗漏了默认模型，首次分离时会转而访问 Hugging Face，并可能触发匿名下载限流；beta.5 已修复这个打包问题。

桌面安装包现在包含 Python 运行时、处理依赖、ffmpeg、yt-dlp、默认 Whisper 小模型和 UVR 人声分离模型。默认流程下载安装后即可使用，不需要 Python、Homebrew、HF Token 或额外模型下载。

选择安装包

下载文件	适用设备	大小
VocalFlow-0.8.0-beta.5-mac-arm64.dmg	Apple Silicon Mac；推荐的原生 K 歌客户端	约 995 MB
VocalFlow.Studio-0.8.0-beta.5-mac-arm64.dmg	Apple Silicon Mac；制作、检查与播放工作台	约 920 MB
VocalFlow.Studio-0.8.0-beta.5-win-x64.exe	64 位 Windows；制作、检查与播放工作台	约 874 MB

目前没有 Intel Mac 安装包。iPhone 版本不包含在本次 GitHub Release。

本版修复

• 正式 Studio macOS / Windows 打包配置现在包含 faster-whisper-small 与 UVR-MDX-NET-Inst_HQ_3.onnx。
• 新增发布后内部检查：Python、FFmpeg、处理脚本或任一默认模型缺失 / 体积异常时，CI 会阻止 Release 发布。
• Separate 阶段改为持续活动状态，不再在无法计算百分比时一直误显示 0%。
• 默认工作流不会访问 Hugging Face；HF Token 仅在用户主动选择可选大模型时需要考虑。

模型、联网与空间

• 内置 faster-whisper-small（约 464–486 MB）和 UVR-MDX-NET-Inst_HQ_3.onnx（约 64 MB）。
• Studio 可选模型首次使用时仍需下载：medium 约 1.53 GB、large-v3-turbo 约 1.62 GB、large-v3 约 3.09 GB。
• 本地文件处理、已导入歌曲包播放和录音可以离线完成；YouTube / 哔哩哔哩媒体下载和在线搜索仍需要网络。
• 建议至少预留 5 GB；使用大模型或保留较多 MV / 中间文件时建议 10 GB 以上。

首次打开

安装包尚未使用 Developer ID 签名和 Apple 公证。macOS 首次启动时，可在 Finder 右键应用选择“打开”，或前往“系统设置 → 隐私与安全性 → 仍要打开”。Windows SmartScreen 出现时，选择“更多信息 → 仍要运行”。

完整使用说明：中文 README · English README

---

English

This is a clean-machine installation fix. VocalFlow Studio beta.4 accidentally omitted the default model files, so first separation fell back to Hugging Face and could hit anonymous-download rate limits. Beta.5 fixes the packaging error.

Desktop installers now contain the Python runtime, processing dependencies, ffmpeg, yt-dlp, the default Whisper small model, and the UVR separator model. The default workflow is ready after installation without Python, Homebrew, an HF Token, or an additional model download.

Choose an installer

Download	Platform	Size
VocalFlow-0.8.0-beta.5-mac-arm64.dmg	Apple Silicon Mac; recommended native karaoke client	About 995 MB
VocalFlow.Studio-0.8.0-beta.5-mac-arm64.dmg	Apple Silicon Mac; creation, review, and playback studio	About 920 MB
VocalFlow.Studio-0.8.0-beta.5-win-x64.exe	64-bit Windows; creation, review, and playback studio	About 874 MB

There is no Intel Mac installer yet. The iPhone client is not included in this GitHub Release.

Fixes

• Production Studio packaging for macOS and Windows now includes faster-whisper-small and UVR-MDX-NET-Inst_HQ_3.onnx.
• CI inspects the packaged app and blocks publishing if Python, FFmpeg, the processing script, or either default model is missing or unexpectedly small.
• The Separate stage now shows continuous indeterminate activity instead of a misleading 0% when fractional progress is unavailable.
• The default workflow does not contact Hugging Face. An HF Token only becomes relevant when the user explicitly selects an optional larger model.

Models, network access, and storage

• Bundled: faster-whisper-small (about 464–486 MB) and UVR-MDX-NET-Inst_HQ_3.onnx (about 64 MB).
• Optional Studio models still download on first use: medium about 1.53 GB, large-v3-turbo about 1.62 GB, and large-v3 about 3.09 GB.
• Local-file processing, imported-package playback, and recording can work offline. YouTube/Bilibili media downloads and online search still require network access.
• Keep at least 5 GB free, or 10 GB or more when using larger models or retaining many MVs and intermediate files.

First launch

The installers are not yet Developer ID signed or Apple-notarized. On macOS, right-click the app in Finder and choose Open, or use System Settings → Privacy & Security → Open Anyway. On Windows, choose More info → Run anyway if SmartScreen appears.

Full documentation: 中文 README · English README

VocalFlow v0.8.0-beta.4

VocalFlow v0.8.0-beta.4

中文 | English

这是一个预发布测试版。桌面安装包已经包含默认处理流程需要的运行时、工具和模型，下载安装后即可使用，不需要另外安装 Python、Homebrew、FFmpeg 或默认模型。

选择安装包

下载文件	适用设备	大小
VocalFlow-0.8.0-beta.4-mac-arm64.dmg	Apple Silicon Mac；推荐的原生 K 歌客户端	约 997 MB
VocalFlow.Studio-0.8.0-beta.4-mac-arm64.dmg	Apple Silicon Mac；制作、检查与播放工作台	约 411 MB
VocalFlow.Studio-0.8.0-beta.4-win-x64.exe	64 位 Windows；制作、检查与播放工作台	约 377 MB

目前没有 Intel Mac 安装包。iPhone 版本仍处于 TestFlight / Xcode 测试流程，本次 Release 不包含 iPhone 安装包。

模型、联网与空间

• 默认内置 faster-whisper-small（约 464–486 MB）和 UVR-MDX-NET-Inst_HQ_3.onnx（约 64 MB），合计约 527 MB，不会在首次处理时重复下载。
• Studio 的“高级”设置可选更大的 Whisper 模型；第一次使用时会从 Hugging Face 下载并缓存在本机：medium 约 1.53 GB、large-v3-turbo 约 1.62 GB、large-v3 约 3.09 GB。
• 本地文件处理、已导入歌曲包播放和录音可以离线完成。媒体链接下载、在线搜索与可选模型首次下载需要网络。
• 建议至少预留 5 GB 可用空间；使用 large-v3 或保留较多 MV / 中间文件时建议预留 10 GB 以上。

首次打开

安装包尚未使用 Developer ID 签名和 Apple 公证。macOS 首次启动时，可在 Finder 右键应用选择“打开”，或前往“系统设置 → 隐私与安全性 → 仍要打开”。如果系统提示应用“已损坏”，确认安装包来自本 Release 后执行：

xattr -dr com.apple.quarantine "/Applications/VocalFlow.app"
# VocalFlow Studio 使用：
xattr -dr com.apple.quarantine "/Applications/VocalFlow Studio.app"

Windows SmartScreen 出现时，选择“更多信息 → 仍要运行”。

本版亮点

• 原生 Mac Room、跨平台 VocalFlow Studio、iPhone 客户端与私有 Mac mini Agent 使用兼容的 K 歌包流程。
• 原生 Mac 与 Studio 均支持演唱录音：麦克风 WAV 原始人声，以及可分享的 M4A / MP3 / WAV 混音。
• Studio 增加录音电平、录音预览、重命名、增益 / 延迟调整和重新导出。
• 自适应 MV、逐字歌词、原唱 / 伴奏切换、队列、YouTube 和哔哩哔哩来源。
• Python 运行时、处理依赖、ffmpeg、yt-dlp 和默认模型随桌面安装包提供。

完整使用说明与语言切换：中文 README · English README

---

English

This is a prerelease. Desktop installers include the runtime, tools, and default models required by the standard workflow. You can use the app after installation without separately installing Python, Homebrew, FFmpeg, or a default model.

Choose an installer

Download	Platform	Size
VocalFlow-0.8.0-beta.4-mac-arm64.dmg	Apple Silicon Mac; recommended native karaoke client	About 997 MB
VocalFlow.Studio-0.8.0-beta.4-mac-arm64.dmg	Apple Silicon Mac; creation, review, and playback studio	About 411 MB
VocalFlow.Studio-0.8.0-beta.4-win-x64.exe	64-bit Windows; creation, review, and playback studio	About 377 MB

There is no Intel Mac installer yet. The iPhone client is still distributed through the TestFlight/Xcode beta workflow and is not included in this Release.

Models, network access, and storage

• faster-whisper-small (about 464–486 MB) and UVR-MDX-NET-Inst_HQ_3.onnx (about 64 MB) are bundled, about 527 MB total. The default workflow does not download them again on first use.
• Larger Whisper models are available in Studio's Advanced settings and download from Hugging Face on first use: medium about 1.53 GB, large-v3-turbo about 1.62 GB, and large-v3 about 3.09 GB.
• Local-file processing, imported-package playback, and recording can work offline. Media URL downloads, online search, and the first optional-model download require network access.
• Keep at least 5 GB free, or 10 GB or more when using large-v3 or retaining many MVs and intermediate files.

First launch

The installers are not yet Developer ID signed or Apple-notarized. On macOS, right-click the app in Finder and choose Open, or use System Settings → Privacy & Security → Open Anyway. If macOS says the app is damaged, first verify that it came from this Release, then run:

xattr -dr com.apple.quarantine "/Applications/VocalFlow.app"
# For VocalFlow Studio:
xattr -dr com.apple.quarantine "/Applications/VocalFlow Studio.app"

On Windows, choose More info → Run anyway if SmartScreen appears.

Highlights

• The native Mac Room, cross-platform VocalFlow Studio, iPhone client, and private Mac mini Agent share a compatible karaoke-package workflow.
• Native Mac and Studio both support performance recording: raw vocal WAV plus a share-ready M4A/MP3/WAV mix.
• Studio adds recording meters, take preview, rename, gain/latency adjustment, and re-export.
• Adaptive MV display, word-synced lyrics, original/backing selection, queue, YouTube, and Bilibili sources.
• Python runtime, processing dependencies, ffmpeg, yt-dlp, and default models ship inside desktop installers.

Full documentation and language switch: 中文 README · English README

VocalFlow v0.8.0-beta.3

VocalFlow 0.8 beta brings the native Mac room, iPhone companion,
private Mac mini Agent, and cross-platform VocalFlow Studio into one
compatible karaoke workflow.

Highlights:

• Immersive native macOS Room with adaptive MV sizing, word-synced lyrics, original/backing mix, queue, search, YouTube and Bilibili sources.
• Performance recording in both native Mac and Electron Room: raw vocal WAV plus a share-ready M4A/MP3/WAV mix.
• iPhone Room can import packages locally or download completed jobs from the private Mac mini Agent.
• Native Mac and iPhone discover both native and Electron package manifests.
• Default faster-whisper-small and UVR-MDX-NET-Inst_HQ_3.onnx weights, Python runtime, ffmpeg, yt-dlp, and processing packages ship inside desktop installers.

Choose one:

• VocalFlow-0.8.0-beta.3-mac-arm64.dmg — recommended native Apple Silicon client.
• VocalFlow.Studio-0.8.0-beta.3-mac-arm64.dmg — Electron studio for macOS.
• VocalFlow.Studio-0.8.0-beta.3-win-x64.exe — Electron studio for Windows.

This is a prerelease. Public macOS distribution still requires the
maintainer's Developer ID certificate and notarization credentials;
iPhone TestFlight upload uses the separate manual workflow.

VocalFlow v0.8.0-beta.2

VocalFlow 0.8 beta brings the native Mac room, iPhone companion,
private Mac mini Agent, and cross-platform VocalFlow Studio into one
compatible karaoke workflow.

Highlights:

• Immersive native macOS Room with adaptive MV sizing, word-synced lyrics, original/backing mix, queue, search, YouTube and Bilibili sources.
• Performance recording in both native Mac and Electron Room: raw vocal WAV plus a share-ready M4A/MP3/WAV mix.
• iPhone Room can import packages locally or download completed jobs from the private Mac mini Agent.
• Native Mac and iPhone discover both native and Electron package manifests.
• Default faster-whisper-small and UVR-MDX-NET-Inst_HQ_3.onnx weights, Python runtime, ffmpeg, yt-dlp, and processing packages ship inside desktop installers.

Choose one:

• VocalFlow-0.8.0-beta.2-mac-arm64.dmg — recommended native Apple Silicon client.
• VocalFlow.Studio-0.8.0-beta.2-mac-arm64.dmg — Electron studio for macOS.
• VocalFlow.Studio-0.8.0-beta.2-win-x64.exe — Electron studio for Windows.

This is a prerelease. Public macOS distribution still requires the
maintainer's Developer ID certificate and notarization credentials;
iPhone TestFlight upload uses the separate manual workflow.

VocalFlow v0.8.0-beta.1

VocalFlow 0.8 beta brings the native Mac room, iPhone companion,
private Mac mini Agent, and cross-platform VocalFlow Studio into one
compatible karaoke workflow.

Highlights:

• Immersive native macOS Room with adaptive MV sizing, word-synced lyrics, original/backing mix, queue, search, YouTube and Bilibili sources.
• Performance recording in both native Mac and Electron Room: raw vocal WAV plus a share-ready M4A/MP3/WAV mix.
• iPhone Room can import packages locally or download completed jobs from the private Mac mini Agent.
• Native Mac and iPhone discover both native and Electron package manifests.
• Default faster-whisper-small and UVR-MDX-NET-Inst_HQ_3.onnx weights, Python runtime, ffmpeg, yt-dlp, and processing packages ship inside desktop installers.

Choose one:

• VocalFlow-0.8.0-mac-arm64.dmg — recommended native Apple Silicon client.
• VocalFlow Studio-0.8.0-beta.1-mac-arm64.dmg — Electron studio for macOS.
• VocalFlow Studio-0.8.0-beta.1-win-x64.exe — Electron studio for Windows.

This is a prerelease. Public macOS distribution still requires the
maintainer's Developer ID certificate and notarization credentials;
iPhone TestFlight upload uses the separate manual workflow.