American tech giant Amazon has unveiled its new AI voice model codenamed Nova Sonic, thus marking a significant step forward in its voice technology offerings.
The goal of Nova Sonic is to create voices that are more expressive, realistic, and natural. Although it improves upon earlier Alexa speech models, it is more realistic and responsive.
According to Amazon, Sonic’s performance on benchmarks evaluating speed, speech recognition, and conversational quality is comparable to that of Google’s and OpenAI’s frontier voice models.
READ ALSO: Amazon submits offer to take over TikToks US operations
Although Amazon hasn’t disclosed all the technical details, Nova Sonic most likely makes use of large-scale neural networks, perhaps with transformer-based architectures, akin to models like Google’s WaveNet and SynthID or OpenAI’s VALL-E.
Alexa+, Amazon’s enhanced digital voice assistant, is powered by Nova Sonic components, according to Rohit Prasad, Amazon SVP and Head Scientist of AGI.
Prasad claimed in an interview that Nova Sonic expands on Amazon’s proficiency with “large orchestration systems,” the technical framework that underpins Alexa. According to Prasad, Nova Sonic is superior to competing AI voice models at directing user requests to various APIs.
© Ripples Nigeria
