Google's Translatotron aims to improve communication by translating tone and cadence as well as words.
As far as real-time translators have come, they still lack the natural flow of human speech. That could one day change with Google’s prototype AI Translatotron translator, which uses an end-to-end system AI model to translate speech directly from audio input to audio output, without first converting it into text. According to the team, the more direct translation reduces errors by eliminating steps and makes it possible to retain the speaker’s natural voice.
More Info about this Invention: