Google’s Voice Generating AI System can Imitate Human Voice
A research paper published by Google in December 2017 highlighted on a text-to-speech system called Tacotron 2, which claims near-human accuracy at imitating audio of a person speaking from text.
Technology giants such as Apple and Google are focused on AI to perform certain actions on electronic devices. Google’s new assistant performs most actions through voice recognition, but it also provides text feedback in a voice that is ever so close to sounding as natural as humans. Voice generation has come a long way from sounding stiff an unnatural to smooth and life-like and a new report suggests Google is closer to achieving the latter.
AI speech from text is so human-like that it is indistinguishable from that of a person reading from text. The system comprises two deep neural networks, the first translates text into a spectrogram (pdf), which is essentially a visual representation of audio frequency. After this, the spectrogram is fed into WaveNet, a speech generation algorithm from Google DeepMind. WaveNet reads the visual to create corresponding audio elements.
Quartz’s report includes a few audio samples where one text sentence is generated by Tacotron 2 and the other is of a human, which is hard to differentiate between the two voice samples. Notably, Tacotron 2 can also handle words that are hard to pronounce as well as names. It is also able to change the way it speaks and stress on certain words depending on the punctuation. The system has only been trained to imitate one voice and Google will have to train it again for it to speak in a different voice.
According to Artificial Intelligence in Telecommunication Market report published by Coherent Market Insights, AI is expected to be more beneficial in telecom industry for communication through Google’s AI, as it has near life-like speech generation, which is more refined as compared to the stilted speech. This technology can be applied to Google’s AI voice service, since WaveNet was deployed for Assistant. This is expected to make assistant more natural, scarily human-like and accurate, bridging the gap between human-computer interactions.
You may be interested
Smart Nanotechnology Enlivens Wooden Ships and ArtifactsMohit Joshi - January 28, 2019
Scientists developed an innovative approach using ‘smart’ nanocomposites to conserve 16th-century British warship, the Mary Rose, and its artifacts. Shipwrecks are preserved in sediments under cold water,…
Researchers Discovered New Properties of Ferroelectric CrystalMohit Joshi - January 25, 2019
Researchers discovered new properties of ferroelectric crystal, according to a study published on July 10, 2018. This study was conducted by a team of researchers from China.…
Researchers Develop Novel Antibody to Knockdown of Protein in VertebratesMohit Joshi - January 24, 2019
Researchers from Technische Universität Dresden developed a synthetic antibody that improves functional analysis of proteins. A research led by Dr. Jörg Mansfeld of the Biotechnology Center of…