Skip to content

OneAsr 1.3.0

Choose a tag to compare

@github-actions github-actions released this 15 Oct 15:05
· 34 commits to master since this release

OneAsr 1.3.0 — hotwords, plus polish

New

  • Hotwords (optional): add the names and terms the model keeps mishearing to a list, and the transcription gets them right. Chinese and English only; the hotword model is a separate download.
  • Text fields: drag to select, undo / redo (Ctrl+Z / Ctrl+Y), a selection highlight you can actually see, and word lists that wrap cleanly.
  • Japanese subtitles: lines no longer break inside a word.

Fixed

  • When hotwords block a task (empty group or missing model), the app opens the hotword page instead of Settings.
  • The word-list box fills the drawer height when the window is maximized; unnamed hotword groups show a localized placeholder instead of machine ids.

macOS Install

The package is unsigned (no Apple Developer account), so the first install needs the system's quarantine mark cleared once; afterwards there are no prompts. The disk image includes 安装说明.txt and 安装 OneAsr.command:

  1. Run it in Terminal — the method that works on every macOS version:
bash "/Volumes/OneAsr 1.3.0/安装 OneAsr.command"
  1. Or right-click 安装 OneAsr.command → "Open" → "Open" again in the dialog.

Download

Click a file name to download:

Platform Download
Windows Installer · Portable
Linux x64 .deb · Portable
macOS (Apple Silicon) Disk image

The first launch downloads the speech-recognition models; everything works offline afterwards.

Full diff: v1.2.5...v1.3.0

OneAsr 1.3.0 — 热词、细节打磨

新增

  • 热词(可选):把总被听错的词放进词表,转写就按你写的出。仅中文与英文;热词模型需另下。
  • 文本输入框:拖选、撤销 / 重做(Ctrl+Z / Ctrl+Y)、看得见的选区高亮,词表折行也更干净。
  • 日语字幕:不再切在词内。

修复

  • 被热词挡住开跑时(组没词 / 模型缺失),界面翻到热词页,而不是设置页。
  • 最大化窗口时词表框长满剩余高度;没起名的热词组显示占位名,而不是 default、g1 这类机器 id。

macOS 安装

包未签名(没有 Apple 开发者账号),首次安装需要解除一次系统隔离标记,装好之后不再提示。磁盘映像里附了「安装说明.txt」和「安装 OneAsr.command」:

  1. 用终端运行 —— 任何 macOS 版本都能用的方式:
bash "/Volumes/OneAsr 1.3.0/安装 OneAsr.command"
  1. 或在访达里右键「安装 OneAsr.command」→ 选「打开」→ 弹框里再点一次「打开」。

下载

点文件名即下载:

平台 下载
Windows 安装包 · 便携包
Linux x64 安装包 · 便携包
macOS(M 芯片) 磁盘映像

首次运行会下载语音识别模型,下完即可离线使用。

完整对比:v1.2.5...v1.3.0