“The upgrade comes at a time of growing excitement over voice recognition technology, especially as voice interaction becomes an increasingly important user interface for mobile devices and other consumer products emerging with the Internet of Things. “
Czech Republic-based Phonexia has announced an upgrade to its Deep Embeddings voice biometrics engine. The new version adds machine learning to the mix, based on a deep neural networks model.
In a statement announcing the upgrade, Deep Embeddings said the machine learning technology has enabled the solution to establish voice templates twice as fast as the previous version of the engine. Its False Accept and False Reject error rates, meanwhile, have dropped by 2.4 times; while the memory requirements take only a seventh of the RAM previously needed by the system.
Elaborating on the potential of the upgraded technology, Phonexia CTO Petr Schwarz asserted that “[i]n addition to making biometric adoption easier for traditional clients, the reduced memory requirements will accelerate adoption of speaker identification into new segments such as 4.0 devices, IoT, and devices with no permanent connection to the Internet.”
The upgrade comes at a time of growing excitement over voice recognition technology, especially as voice interaction becomes an increasingly important user interface for mobile devices and other consumer products emerging with the Internet of Things. At the same time, there is considerable buzz over the kind of machine learning technology that powers Deep Embeddings, which should help Phonexia to attract some attention as it markets its enhanced solution.
0 comments:
Post a Comment