Skip to content

Releases: Longt-audio/voxminutes

VoxMinutes v0.1.2

Choose a tag to compare

@Longt-audio Longt-audio released this 29 Jul 14:32

VoxMinutes v0.1.2 — download links everywhere you need them.

What's new

  • Links panel everywhere: the per-source download links (GitHub / HuggingFace / hf-mirror / ModelScope) are now available on all model rows in Settings — including already-installed ones — and inside the onboarding wizard. Pick a source, or copy links to a download manager and install via Import.
  • See what's downloading: the progress line now shows the host of the URL being downloaded (e.g. hf-mirror.com), so you can tell which mirror is in use.
  • Footer version now reflects the actual app version.

Install notes

  • No models are bundled. On first launch the onboarding wizard guides you to download or import models. Mainland China works out of the box (automatic mirror fallback).
  • Windows may show a SmartScreen warning because the installer is not code-signed — click More info → Run anyway.

中文说明:现在设置页所有模型行(含已安装)和首启向导里都有「链接」面板,可换源下载或复制到下载器后用「导入」安装;下载进度会显示当前使用的下载源域名。蓝色 SmartScreen 提示点"更多信息 → 仍要运行"即可。

Supersedes v0.1.1. Full details in the README.

VoxMinutes v0.1.1

Choose a tag to compare

@Longt-audio Longt-audio released this 28 Jul 15:31

VoxMinutes v0.1.1 — easier, more reliable model downloads.

What's new

  • Pick your download source: in Settings → Models, each model now has a "Links" panel listing all mirrors (GitHub / HuggingFace / hf-mirror / ModelScope). Download from a specific source, or copy the links into your favorite download manager (IDM, aria2, etc.) and install via Import.
  • README: manual download links for every model in English / 中文 / 한국어 / 日本語.
  • Downloads now resume from where they stopped (HTTP Range), and multiple models can download in parallel.

Install notes

  • No models are bundled. On first launch the onboarding wizard guides you to download or import models. Mainland China works out of the box (automatic mirror fallback).
  • Windows may show a SmartScreen warning because the installer is not code-signed — click More info → Run anyway.

中文说明:本次更新让每个模型都能选择/复制下载源(设置 → 模型 → 链接),下载慢时可改用镜像或复制到下载器后用「导入」安装。安装包不内置模型,首启向导会引导下载。蓝色 SmartScreen 提示点"更多信息 → 仍要运行"即可。

Supersedes v0.1.0. Full details in the README.

VoxMinutes v0.1.0

Choose a tag to compare

@Longt-audio Longt-audio released this 26 Jul 15:19

VoxMinutes v0.1.0 — first public MVP release.

Your local meeting assistant: records system audio & mic together, real-time transcription, translation and summaries — all on your device.

Highlights

  • Real-time streaming transcription (X-ASR / SenseVoice)
  • Dual-channel recording: system playback + microphone
  • Live translation: OPUS-MT (fast) or Hy-MT2 (13 target languages)
  • Local AI meeting summaries (Qwen / Gemma GGUF, on-device)
  • History, search, inline editing, TXT / SRT / Markdown export
  • Model manager: multi-source downloads (GitHub / hf-mirror / ModelScope) with resume, or import local archives / GGUF
  • First-launch onboarding wizard, UI in English / 中文 / 한국어 / 日本語

Install notes

  • No models are bundled. On first launch the onboarding wizard guides you to download or import models. Mainland China works out of the box (automatic mirror fallback).
  • Windows may show a SmartScreen warning ("Windows protected your PC") because the installer is not code-signed — click More info → Run anyway.
  • Requirements: Windows 10/11 x64. All models run on CPU; 8 GB RAM recommended (16 GB if using a local summary model).

中文说明:安装包不内置模型,首次启动会弹出新手指引帮你下载/导入模型(国内网络可直连镜像)。出现蓝色 SmartScreen 提示时点"更多信息 → 仍要运行"即可。

Full details: see the README (中文).