Your Apps Can Talk! Introducing Cloud Text-to-Speech, Powered by WaveNet Technology
summary
Cloud Text-to-Speech brings WaveNet quality to an API claiming 70% closer to human speech than traditional synthesis, with a Cloud TTS pay-as-you-go service that supports 30+ languages and 50+ voices. The system uses WaveNet distilled into Parallel WaveNet (about 3 orders of magnitude faster, practical ~20x real time) and runs on TPUs; features include audio profiles tuned for handset, home or phone lines. Shown production pieces use the Cloud Text-to-Speech API plus Cloud Storage, Cloud Functions, Cloud SQL and DialogFlow to generate, store and serve samples for IVR, IoT and media.
AI-written summary · about
chapters (27)
- 0:00Introduction
- 0:41Previous Approaches
- 1:38Limitations
- 3:25WaveNet
- 4:43Demo
- 6:58What is WaveNet
- 10:02Parallel WaveNet
- 11:25WaveNet Speed Increase
- 11:50Naturalness Gap
- 12:39Benefits of Cloud TTS
- 14:57What is Cloud TTS
- 15:34WaveNet Languages
- 16:34Audio Profiles
- 17:42Audio Profiles Example
- 18:35TexttoSpeech Demo
- 21:34TexttoSpeech Example
- 29:14Translate Text Example
- 35:23Ingram Content Group
- 36:23Three Key Themes
- 36:45Our Innovation
- 37:13Barriers to Entry
- 37:51Business Trends
- 39:56Accessibility
- 41:09Book Discoverability
- 45:09Core Source
- 46:24Secret Sauce
- 47:15Conclusion
related talks
17mServing Voice AI at Scale
01h 00mBoost your App with Machine Learning APIs
58mAdding Speech to your Web App_UIArchitecture
26mEnabling the next generation in voice technology
46mArtificial intelligence to the rescue of accessibility
47mBuilding smarter apps with Machine Learning, from magic to reality
30mTalk to Your App
36mVoice Enabling Mobile Apps and Custom Hardware Devices