README file from
GithubVoice Notes 语音笔记
Listen to your Obsidian notes. Read full text or AI-summarized versions aloud — hands-free, eyes-free.
用语音朗读你的 Obsidian 笔记。支持全文朗读和大模型概括后朗读，解放双眼，随时随地听笔记。
Why Note Reader / 为什么用笔记朗读
Ever finished writing a note and wanted to just... lean back and listen to it? Voice Notes turns your vault into a personal podcast. Powered by Microsoft Edge TTS for natural-sounding speech and optional LLM summarization, it's built for people who learn better by listening — or just want to rest their eyes after a long writing session.
写完笔记想闭眼听一遍？语音笔记把你的笔记库变成私人播客。基于微软 Edge TTS 自然语音引擎，可选大模型概括，适合习惯听觉学习的人，也适合写完笔记后放松眼睛听一遍回顾。
Features / 功能
Text-to-Speech / 语音朗读
- Full-note reading: Reads the entire note from top to bottom
- AI-summarized reading: Sends the note to an LLM for summarization, then reads the summary aloud — great for quickly grasping the key points of long notes
- 5 compression levels (1%–50%): From one-sentence takeaway to half-length digest
- 全文朗读：从头到尾朗读完整笔记
- 概括朗读：先调用大模型生成摘要再朗读，快速抓住长文要点
- 5 档压缩比例（1%–50%）：从一句话总结到半文摘
Smart Playback / 智能播放
- Auto-next: Automatically opens and reads the next note when the current one finishes — binge-listen your entire vault
- Stop after current: Stops after one note
- Loop current: Repeats the same note
- Pause / Resume / Stop: Full playback control at any time
- Progress bar: Real-time progress indicator with elapsed/total time display
- Segment text view: Shows the full text of the current segment with scrollbar
- 自动读下一篇：当前笔记读完后自动打开下一篇继续朗读，一口气听完整个笔记库
- 读完即停：读完当前笔记就停止
- 循环播当前：反复朗读当前笔记
- 暂停 / 继续 / 停止：随时控制播放
- 进度条：实时显示当前段播放进度和时间
- 段落文字：显示当前朗读段的完整文字，带滚动条
Markdown-Aware / Markdown 智能解析
Tables, images, code blocks, math formulas, task lists, and links are automatically converted to spoken-language-friendly text. No more hearing | and # read aloud.
表格、图片、代码块、数学公式、任务列表、链接等自动转换为适合朗读的文本，不会再听到 | 和 # 等符号。
LLM Integration / 大模型集成
Works with any OpenAI-compatible API — OpenAI, DeepSeek, Qwen (通义), Zhipu (智谱), and more. The summary prompt is tuned to produce clean, conversational output ready for speech synthesis, with meta-text and thinking process automatically stripped.
兼容任何 OpenAI 接口格式的服务——OpenAI、DeepSeek、通义千问、智谱等均可。概括提示词针对语音合成优化，自动去除思考过程和元文本，输出干净自然的口语化摘要。
Voice Options / 语音选项
- 20+ Chinese neural voices (male & female, various styles)
- Adjustable speed (0.5x–2.0x) and volume
- 20+ 种中文神经语音（男女多风格可选）
- 语速（0.5–2.0 倍）和音量可调
Getting Started / 快速开始
-
Open a note in Obsidian
-
Click the headphones icon in the left ribbon, or use the command palette:
- Read current note (full) — reads the entire note
- Summarize and read current note — summarizes then reads
-
A player panel appears at the bottom with playback controls
-
在 Obsidian 中打开一篇笔记
-
点击左侧栏的耳机图标，或使用命令面板：
- 朗读当前笔记（全部） — 全文朗读
- 概括后朗读当前笔记 — 先概括再朗读
-
底部出现播放面板，可暂停、继续、停止
Configuration / 配置
Settings > Community plugins > Voice Notes
| Setting / 设置 | Description / 说明 |
|---|---|
| API Base URL | OpenAI-compatible endpoint / OpenAI 兼容接口地址 |
| API Key | LLM API key / 大模型密钥 |
| Model | Model name for summarization / 概括用的模型 |
| Summary ratio | Compression level 1%–50% / 压缩比例 1%–50% |
| Playback mode | Auto-next / Stop / Loop / 自动读下一篇 / 读完即停 / 循环播当前 |
| Voice | TTS voice / 朗读音色 |
| Speed | 0.5–2.0x / 语速 |
| Volume | 0–1 / 音量 |
TTS is powered by Microsoft Edge TTS (msedge-tts) — no API key or additional setup required for speech synthesis.
语音合成基于微软 Edge TTS，无需额外申请密钥或配置。
Technical Details / 技术说明
- TTS engine: Microsoft Edge TTS via WebSocket, supports custom Origin headers for Obsidian's Electron environment
- LLM integration: OpenAI-compatible Chat Completions API, 60s timeout with friendly error messages
- Audio playback: Segment-based synthesis (max 500 chars per segment) with automatic progression
- Memory safe: Object URLs are properly revoked to prevent leaks
- Desktop only: Requires Node.js WebSocket implementation (isDesktopOnly)