Core Features
- True Local TTS: Uses the Microsoft Edge browser engine, no API required
- Zero-dependency Deployment: Only requires installing the edge-tts library
- Fully Offline: No network connection required after the first download
- Multilingual Support: Chinese, English, Japanese, Korean, etc.
- Multiple Voices: 20+ preset voices, covering male, female, and child voices
Use Cases
- Video dubbing: short videos, documentaries, corporate promotional videos
- Audiobook production: novels, stories, study materials
- Assisted reading: webpage reading, document reading
- Multilingual learning: foreign language pronunciation practice
- Offline use: speech synthesis in no-network environments
Trigger Method
Make a request directly, no configuration required.
"朗读这段文本，用温柔女声"
"生成英语配音"
"用日语朗读这段话"
"用男声朗读解说词"
Workflow
The service automatically performs:
- Text analysis: identify language and emotional tone
- Voice selection: automatically select an appropriate preset voice
- Speech generation: call the local Edge-TTS engine
- Result verification: verify audio quality and integrity
Voice List
Chinese Voices
| Voice Name | Gender | Style | Use Cases |
|---|---|---|---|
| zh-CN-XiaoxiaoNeural | Female | Young and lively | Advertising, short videos |
| zh-CN-YunyangNeural | Male | Calm | Documentaries, corporate promotion |
| zh-CN-XiaohanNeural | Female | Intellectual | News broadcasting |
| zh-CN-YunjianNeural | Male | Deep | Movie narration |
| zh-CN-XiaomengNeural | Female | Gentle | Audiobooks, stories |
| zh-CN-YunxiNeural | Male | Lively | Game commentary |
English Voices
| Voice Name | Gender | Style | Use Cases |
|---|---|---|---|
| en-US-JennyNeural | Female | American | Business dialogue |
| en-US-GuyNeural | Male | American | News broadcasting |
| en-GB-SoniaNeural | Female | British | Formal occasions |
| en-GB-RyanNeural | Male | British | Literary reading |
Other Languages
- Japanese: ja-JP-NanamiNeural (Female), ja-JP-KeitaNeural (Male)
- Korean: ko-KR-SunHiNeural (Female), ko-KR-InJoonNeural (Male)
- German: de-DE-KatjaNeural (Female), de-DE-ConradNeural (Male)
- French: fr-FR-DeniseNeural (Female), fr-FR-HenriNeural (Male)
Usage Examples
Example 1: Basic Text-to-Speech
"朗读这段文本：欢迎使用本地语音合成服务"
Example 2: Specify Voice
"用温柔女声朗读这段文本：今天天气真好"
Example 3: Multilingual
"用英语朗读这段话：Hello, how are you?"
"用日语朗读这段话：こんにちは"
Example 4: Emotional Expression
"用激昂的语调朗读这段解说词：让我们勇往直前！"
"用温柔的语调朗读这段故事：很久很久以前..."
Resource Index
- TTS Generation Script: see scripts/tts_generator.py
- Audio Processing Script: see scripts/audio_processor.py
- Voice Parameter Guide: see references/voice-guide.md
Notes
- Zero Dependencies: only the edge-tts library needs to be installed
- Offline Use: no network required after the first download
- Voice Limitations: uses Microsoft Edge preset voices, custom cloning is not supported
- Emotion Control: emotional variation is achieved by selecting different voices
- Best Experience: the voice package is downloaded on first use, about 50-100MB
Installation Instructions
# 安装 edge-tts
pip install edge-tts
# 验证安装
edge-tts --help
Technical Description
How Edge-TTS Works：
- Uses the online TTS service of the Microsoft Edge browser
- API interface obtained through reverse engineering
- Completely free, no account registration required
- Audio is cached locally after the first download
Differences from API Calls：
- ❌ No API_KEY required
- ❌ No account registration required
- ✅ Offline caching
- ✅ Zero-cost use