Google trains AI to speak more clearly and naturally.

Google will update the Speech Services text-to-speech engine on Android devices, delivering voices that are clearer and more natural-sounding.

See more

Text-to-speech is getting a voice upgrade 🗣️ 👂

The automatic Speech Services by Google update covers 421 voices in 67 languages to offer a more natural-sounding speaker on all 64-bit Android devices 🎉

Hear the difference ↓ https://t.co/LJw9faGMJC

— Android Developers (@AndroidDev) September 28, 2022

According to the developers, users won’t have to do anything—the upgrade will occur behind the scenes. The update will significantly improve the quality of generated speech, particularly in terms of clarity and naturalness, they added.

https://forklog.com/wp-content/uploads/sfg_old.wav	https://forklog.com/wp-content/uploads/new_iog.wav

The 421 voices across 67 languages will receive a new voice model and synthesizer. The current American English voice will automatically be updated to a speech generated with ‘more up-to-date data’.

The developers also showed samples of updated voices in other languages.

https://forklog.com/wp-content/uploads/afs_old.wav	https://forklog.com/wp-content/uploads/afs_new.wav
https://forklog.com/wp-content/uploads/esf_old.wav	https://forklog.com/wp-content/uploads/esf_new.wav

The company will roll out the Speech Services update to all 64-bit Android devices via the Google Play Store over the coming weeks.

Earlier in September, OpenAI unveiled the open-source Whisper speech-recognition system, capable of transcription in multiple languages.

In August, the streaming service Megogo employed AI to provide AI-generated voiceovers for video content.

In May 2021, Google unveiled the neural network model LaMDA, which communicates like a living human and supports casual dialogue on a range of topics.

Subscribe to ForkLog news on Telegram: ForkLog AI — all the news from the world of AI!