Google’s new Translatotron AI will do speech-to-speech translation

Google has designed a strong new unnaturally intelligent system that’s capable of translating the word from one language to the opposite, while not the requirement for text.

Called Translatotron, the new system relies on a sequence-to-sequence network that interprets speech victimisation supply spectrograms as input and generates spectrograms of the translated content within the target language.

The highlight of the planned system is that it’s able to retain the vocal characteristics of the first speaker within the translated speech victimisation speaker encoder network, creating it sound additional natural.(FA9NEWS)

Leave a Reply

Your email address will not be published. Required fields are marked *