Hybrid TTS Update History

Hybrid TTS is an advanced feature to enable embed devices with cloud AI model. The Hybrid Speech SDK build in online connectivity, offline models, and hybrid arbitration logics. The product improved mobile, IOT, smart car experience with leading-edge AI technology and solutions.

2021.11 Hybrid TTS Updates

• Improved offline model CPU throttling feature
• Updated the pronunciation dictionary, fixed 127 car domain related issues
• Added a new zh-CN male voice hybrid model (Yunxi)
• Optimized the online mix-lingual pronunciation quality
• Optimized the online model word breaking, emphasis, prosody

2021.11 混合云语音合成产品更新

• 增强离线模型的CPU限速控制功能
• 更新中文发音字典，修正127个车载场景发音问题
• 新增一个中文男声混合云模型(云希)
• 优化在线模型双语混读发音质量
• 优化在线模型断句、重音、语调发音质量

2021.10 Hybrid TTS Updates

• Optimized the zh-CN POI pronunciation in Beijing, Shanghai, Guangzhou
• Improved the zh-CN polyphony and person name reading
• Improved the cache algorithm with background updating feature
• Introduced a new codec for lower bandwidth and lower latency
• Reduced the TTS initial time in some conditions
• Fixed a rare crash issue with special Unicode chars
• Fixed a TTS service random stop issue after repeatedly stop operations

2021.10 混合云语音合成产品更新

• 优化中文北京、上海、广州POI发音规则
• 提升中文多音字，人名发音准确率
• 提升缓存算法，加入后台线程更新功能
• 引入最新的音频编码，实现更低带宽更低延迟
• 减少语音合成模块在特定条件下的启动时间
• 修复极个别情况下，文本带有特殊Unicode字符造成的崩溃
• 修复在反复终止操作时，语音合成随机性服务停止的问题