LexVoice 2.5.0
LexVoice 2.5.0
Released: 2026-09-15
Changes since 2.4.0.
English and Japanese support
- One plugin with Chinese, English, and Japanese interfaces. LexVoice follows the language selected in Obsidian, including settings, menus, suggested questions, knowledge actions, and processing controls.
- Built-in AI note templates follow Obsidian's language. The default AI output policy also follows the host; existing explicit language preferences and custom prompts are preserved.
- Newly generated recording headings, segment labels, and unchanged built-in daily-note templates follow the host language. Known built-in labels on older derived notes are localized for display without rewriting their files.
- Existing notes, raw transcripts, custom template names, paths, and historical error messages are not automatically translated. Speech recognition retains the spoken language rather than translating audio to the interface language.
Layout fixes
- Long menu labels wrap and stay within the window, including audio-input choices.
- Recording input status and level bars have separate space, preventing text overlap.
- Speaker-count and name inputs remain readable in narrow import dialogs.
- Fixed overlapping setup-state buttons, processing controls, and crowded library labels.
Updating
Update through LexVoice's update settings, or download main.js, manifest.json, and styles.css from this release. Finish any active recording or processing first, then reload the plugin. Keep data.json, recordings, and notes.
Only distribution artifacts and user documentation are public. GitHub's automatically generated "Source code" archives contain the public distribution snapshot, not private development source.
中文说明
相对 2.4.0，本版新增中英日界面及内置 AI 模板语言跟随，覆盖设置、菜单、建议问题、知识整理及处理控件。默认 AI 输出语言跟随 Obsidian，已有明确语言偏好和自定义提示词保留。
新录音标题、分段标签和未修改的内置日记模板随宿主语言生成；已知内置派生标签只转换显示，不改写文件。旧笔记、原始转写、自定义名称、路径和历史错误不自动翻译，原始转写保留实际发言语言。
修复长菜单截断、状态按钮重叠、录音电平条遮挡文字、说话人输入框挤压和资料卡布局。更新前请先完成录音和处理任务，保留配置、录音及笔记，再重载插件。