Knowles and Fluent.ai Partner to Make Offline Voice-Controlled Hearables
Speech recognition tech Fluent.ai and audio hardware maker Knowles have teamed up to create True Wireless Stereo (TWS) earbuds and hearables that can process voice interactions on-edge, meaning without internet access. Hearables built with their system will allow wearers to talk with a voice assistant and carry out on-device tasks without needing to reach a cloud server as a way of boosting speed and privacy for users.
Fluent is known for its embedded speech recognition software that can be embedded in devices. The company’s speech-to-intent technology can handle a range of voice commands related to phone calls, music, and other smart device operations quickly and within a relatively small physical device. That makes a natural fit with Knowles and its micro-acoustic microphones, speakers, and audio processing equipment. Fluent’s services can be embedded in the Knowles AISonic Audio Edge Processor IA8201, which runs entirely within a device. That makes it faster at responding to voice commands and avoids any concerns about data storage within the cloud.
“We are honored to be partnering with Knowles, a global leader in audio solutions innovating in the hearables space,” Fluent CEO Probal Lala said. “TWS use cases fit Fluent’s unmatched advantages perfectly as hearables are intended for consumer use anytime and anywhere. Our embedded technology enables users of any language or accent to use their earbuds off the grid and in noisy environments without needing to worry about noise levels, battery life, or privacy. This technology presents hearable OEMs the opportunity to provide multilingual command access to a truly global customer base for the first time.”
Interest in on-device processing for voice AI has risen to become a frequent element for third-party providers. Speech tech developer Sensory offers a customizable voice assistant specifically for smart home appliances with no need for the cloud. Sensory also has its VoiceHub platform to streamline creating custom wake words for those devices. The same features are part of the Picovoice AI platform and ID R&D’s vocal identification engine, which adds vocal security to devices without needing to transmit data. The tech giants aren’t automatically opposed to the on-edge systems. Many Alexa or Google Assistant-enabled devices include on-edge elements like the analog wake word identifier from Aspinity, Sensory’s TrulyHandsfree speech recognition software for iOS and Android apps. That also includes Knowles’ own Amazon-approved Alexa headset development kit.
“Implementing robust, natural sounding embedded voice control in TWS devices has been a challenge due to the memory and compute limitations of current Bluetooth SoCs, and the low power consumption requirements in TWS devices,” Knowles strategic marketing director Raj Senguttuvan said, “Fluent.ai’s small-footprint AI software solution running on Knowles IA8201 can be integrated into TWS earbuds for accurate voice command recognition without the need for an accompanying smartphone app or cloud connectivity. Our joint solution offers hearables OEMs a platform to develop high-quality and flexible voice control in TWS devices for a natural and convenient user experience.”