Listen and try to distinguish Google's new AI-generated voice from human voices

Tacotron 2, Google's artificial intelligence system can turn innocent words into human-like voices. This is considered a great step forward of Google, helping AI become closer to users.

To do so, Tacotron 2 uses two different artificial intelligence roles, the first one will focus on transforming the word into a universal sound (spectrogram - sound chart), the second one (WaveNet ) will turn that data into sound.

Listen and try to distinguish Google's new AI-generated voice from human voices Picture 1

Experience the superior power of Google's new AI system by listening to pairs of identical audio files created by humans and Tacotron 2. Can you tell whether it is a machine or a human voice?

'George Washington was the first President of the United States. "

"That girl did a video about Star Wars lipstick."

"Phần mềm không có vấn đề, chúng thể cung cấp một Solution."

"The buses aren't the PROBLEM, they really provide a SOLUTION."

"Generative adversarial network or variational auto-encoder."

"Basilar membrane and otolaryngology are not auto-correlations."

"He has read the whole thing."

"He reads books".

According to the current announcement, Tacotron 2 can handle words that are difficult to pronounce that people have never met before, can analyze to find the focus to be pressed in the sentence, emphasize what is written flowers or self-analysis to detect basic misspellings.

The Tacotron 2 system currently only produces a single female voice, speaks English and is used to communicate with the virtual assistant Google Assistant. Certainly later, Google will create more voices and Tacotron 2 will be applied to many different platforms.

See more:

1. AI creates a fake video that a person is talking like
2. Can people and robots give birth together in 100 years?
3. The AI ​​system can predict a stroke risk more accurately than a doctor