Google’s team managed to do the ”impossible”: translate your speech in your almost exact voice.
It’s not perfect of course, but the differences between the original audio and the resulting one are smaller when compared to the work of other translation engines.
Google’s AI translator works much simpler than before. It directly converts the audio input to the audio output without in-between steps like traditional systems have.
Instead, Google has created an end-to-end speech-to-speech translation system called the ‘Translatotron’. The system starts with audio spectrograms from input languages into output ones trained to map each other, converts them into audio waves only to have the voice of the original speaker come out at the end.
That’s all it takes! This much more streamlined process allows for fewer mistakes and can change the way audio translations are done.
Follow TechTheLead on Google News to get the news first.
