American technology company Amazon has developed an artificial intelligence (AI) model to convert text into synthesized speech. According to the creators, the neural network has become the largest system of this type ever created. The research results are published at: portal scientific publications arXiv.
The model, called Massive Adaptive Streaming TTS with Immediate Options (BASE TTS), has 980 million parameters and was trained using 100,000 hours of recorded speech samples, mostly in English.
The team also provided examples of the pronunciation of words and phrases in other languages so that the model could correctly pronounce “adios, amigo” and other familiar phrases.
Developers tested BASE TTS on small datasets. It turns out that artificial intelligence can use complex nouns, express emotions and use punctuation marks, as well as ask questions by highlighting the right words.
Amazon plans to use BASE TTS for educational purposes as a learning application.
Formerly Apple developed AI tool for creating animations.