BabbleLabs Launches Clear Edge for Developers
BabbleLabs, a speech science technology provider, has released Clear Edge for Developers, a software development kit that provides instant access to noise-elimination technology.
When integrated into their audio stacks, VoIP providers and conferencing, collaboration and communications software vendors can build edge software applications with artificial intelligence-based speech enhancement.
Through the combination of deep learning, speech science, and embedded systems, Clear Edge technology allows software vendors to provide productive calls from anywhere, on any device – even in the noisiest environments. It reduces noise by as much as 40 decibels. The software allows application vendors to balance compute level, power and speech output clarity to fit every use case across devices and applications.
"Every organization is challenged with innovating quickly," said Samer Hijazi, chief technology officer and co-founder of BabbleLabs, in a statement. "With limited CPU, memory, and power footprint constraints, building AI technology that works on edge devices is complex. Clear Edge for Developers allows application vendors and their audio software engineering teams to accelerate innovation, leveraging proven technology to deliver an outstanding customer experience faster and with minimal investment."
The kit is available as a code library for macOS, Windows, Android, iOS, and Linux and runs on standard x86 and ARM processors.
Cisco will add BabbleLabs' noise cancellation and speech enhancement technologies to its collaboration products.
BabbleLabs, an emerging start-up changing communication between humans and machines, introduced the BabbleLabs Clear product family to enhance speech quality, accuracy, and personalization. Accessible through an easy to use, single-line cloud API, the Clear Cloud product delivers speech enhancement technology driven by a powerful combination of digital signal processing (DSP) and deep neural networks (DNN) trained with a massive specialized data set.