Audio Inbox v2.1.1

v2.1.1 — Memo Mode 📝

✨ New Feature: AI-Powered Memo Classification

Audio Inbox now intelligently distinguishes between reminders and memos, saving each to a dedicated file.

How it works

The AI analyzes each recording and classifies it as:

• 📌 提醒事项 (Reminder) — has actionable tasks → saved to 待办事项.md
• 💭 备忘录 (Memo) — records info/ideas/knowledge → saved to new 备忘录.md
• 🔀 混合 (Mixed) — has both → saved to both files

Memo file format

Each memo entry includes:

• 📝 AI 总结 — AI-organized, cleaned-up version
• 🗣️ 原话 — the original verbatim transcript from STT
• 📁 Link to the audio recording

🔧 Improvements

• Smarter AI prompt with structured output (### 类型 / ### 总结 / ### 待办事项 / ### 备忘内容)
• Robust parser handles both new and legacy AI response formats
• Fallback type inference when AI doesn't output a type section
• Better console logging for debugging
• Auto-migration of old prompts to new format

📦 Full Changelog

See CHANGELOG.md for details.

Audio Inbox v2.0.5

v2.0.5 - Final review compliance fixes

Fixed

• Static styles: Use CSS class .ai-fab-dragged instead of setProperty() for right/bottom reset
• minAppVersion: Bump to 1.4.0 to cover vault.createFolder() and all other APIs in use
• Settings heading: Remove plugin name from settings heading (Obsidian already shows it)
• globalThis: Replace typeof globalThis with safer type cast for popout compatibility

This version addresses all remaining errors from the automated scanner.

Audio Inbox v2.0.4

v2.0.4 - Full community review compliance

Fixed (Errors)

• Use Platform.isMobileApp instead of navigator.userAgent for OS detection
• Use setCssProps() instead of direct style assignment for FAB positioning
• Bump minAppVersion to 1.2.0 to match APIs in use (activeDocument, setHeading, etc.)
• Use new Setting().setName().setHeading() instead of createEl("h3")

Fixed (Warnings)

• Use window.setInterval / window.clearInterval / window.setTimeout
• Use activeDocument instead of document
• Remove all any types with proper TypeScript types
• Remove unused variables (dot, e)
• Mark fire-and-forget promises with void operator
• Remove Node.js builtin module imports (fs, path, os)
• Remove require() calls
• Remove builtin-modules from package.json

Changed

• saveToShortcutsFolder() now uses clipboard API only (was: desktop fs + mobile clipboard)
• markTodosDone() no longer touches filesystem outside vault

This version should pass all automated and manual community review checks.

v2.0.1 — 修复手机端待办不更新

🐛 关键修复

手机端 不更新

根因： 使用了 Node.js 模块，这在手机端（Capacitor/Cordova）不存在，导致手机录音后待办无法同步到 Shortcuts 文件夹。

修复方案：

• 桌面端：行为不变，继续用 Node.js fs 写入 Shortcuts iCloud 文件夹
• 手机端：自动检测平台，切换到剪贴板回退方案

新增

• 平台检测方法
• 添加 console 日志
• 手机端剪贴板回退 + 通知提示

升级方法

重新下载 main.js、manifest.json、styles.css 覆盖插件目录即可。

v2.0.0 — 首个公开版本

🎉 Audio Inbox v2.0.0

✨ 核心功能

功能	说明
🎙️ 一键录音	点击麦克风 → 说话 → 停止 → 完成
🧠 智能转写	SiliconFlow SenseVoiceSmall（免费，国内可用）
🤖 AI 总结	DeepSeek 自动生成结构化笔记 + 待办
📱 手机支持	可拖动悬浮按钮，iOS/Android 通用
✅ 待办提取	自动提取到独立文件
🍎 提醒事项同步	iOS 快捷指令一键同步
📂 批量处理	处理已有音频文件
🌐 国内友好	无需 VPN

🔧 命令

• 开始语音笔记（录音）
• 处理录音文件夹
• 标记已同步的待办为完成

🐛 修复

• STT 返回空文本 → WAV 自动转换
• CORS 错误 → requestUrl
• 手机按钮触摸冲突 → 事件委托
• 待办重复添加 → 覆盖写入

📦 安装

1. 下载 main.js、manifest.json、styles.css
2. 放入 vault/.obsidian/plugins/audio-inbox/
3. 重启 Obsidian → 启用插件