Skip to content

Google's $500 Million Learning Investment Pays Off With an AI That Mimics Human Speech

Google's $500 Million Learning Investment Pays Off With an AI That Mimics Human Speech

There are many speech generation programs that create artificial human speech such as Minitalk and eSpeak, but Google believes they have a system that outperforms current technology by 50%. Deepmind is a Google unit that is working on super-intelligent computers and they’ve created an artificial intelligence that can closely mimic human speech. The AI, called WaveNet, works by analyzing a human voice’s actual sound waves.

More archaic speech generators either use short clips of a previously recorded speaker or electronically generate speech based on how certain letter combinations are supposed to be pronounced. The results of both provide highly accurate speech, but they sound robotic and lack the fluidity of human diction. WaveNet is analyzing actual human speech so it is leaps forward when recreating them. While it outperformed Google’s current text-to-speech programs, it does not yet quite match real human speech — but it’s getting closer. So, no Terminator/Skynet concerns just yet.

Unfortunately, there’s a significant drawback to widespread utilization of WaveNet in it’s current form: Computational power. Older speech generators utilized data sets of short clips to create their language, but WaveNet has to sample the training audio signal 16,000 or more times and then has to form a prediction for the next soundwave based on the previous sample. Bloomberg Technology notes that, despite this steep challenge, tech companies will keep a close eye on WaveNet’s evolution due to the increasing importance of how we all interact with our personal technology. For example, Amazon has its Echo household device that plays music, answers questions, and taps into other smart devices via voice command. Devices such as that would benefit greatly from more human, conversational voices to interact with.

UK’s Deepmind was acquired by Google for about $533 million in early 2014 as a response to deep learning initiatives by Yahoo, IBM, and others, which could have big implications for AI in VR or even AR experiences. One of Deepmind’s notable accomplishments since the acquisition is creating the AI AlphaGo, which defeated the top ranked player of the ancient Chinese strategy game, Go.

Featured Image: BidnessEtc

UploadVR Member Takes

Weekly Newsletter

See More