Using machine learning, CereProc’s CereWave AI V6.1 system sounds indistinguishable from human speech, but it also allows customers to run neural TTS voices on mobile devices, a technology currently not being offered by any other TTS provider.
“This is a major step forward for speech synthesis technology and allows users who have security and privacy concerns to access CereProc’s neural voices locally on-device, preventing sensitive information from being sent to and from web-based servers,” explained Dr Matthew Aylett, Chief Scientific Officer at CereProc.
“CereWave AI 6.1 supports 24khz for increased clarity, naturalness in the linguistic intonation of CereProc’s voices and allows CereProc to create the most customisable and expressive voices available with as little as 3 hours of data.”
CereWave AI 6.1 also offers a machine learning powered editing suite, enabling complete control over CereProc’s neural voices which allows for modification of rate, pitch, amplitude, and emotional expressions. This is a major feature for markets such as healthcare, consumer electronics, retail, transport and large enterprises, where expanded synthetic speech functionality is increasingly in demand, as it gives customers the ability to customise genres for custom voices that reflect the needs of their individual businesses.
For example, an app for healthcare may require a custom voice with an empathic genre to create trust with users. Healthcare, alongside automotive, finance and education, are accelerating the growth of TTS - forecasts from Technavio are predicting the TTS market will be worth $1.76bn by 2022.
“This is a game changing moment for the speech synthesis industry and the billion-dollar Text-to-Speech market. We have managed to overcome one of the most notable limitations that has challenged the major vendors for years. It’s a huge step forwards for both us as a company and the wider industry, ” commented Paul Welham, Executive Chairman of CereProc. “This offering is particularly attractive to customers in any vertical market or system integrators/application developers, who want to their users to access high quality neural voices anywhere, securely.”