Near-Zero-Power Voice-Activation for Battery-Powered Devices

Vesper, DSP Group and Sensory have demonstrated a turnkey development platform that boasts the lowest overall power consumption for far-field always-listening voice interfaces. The DSP Group also recently unveiled its new DBMD5 audio SoC built to drive clearer human-machine voice interactions.Vesper Sensory

The Vesper-DSP Group-Sensory integrated development platform features Sensory’s voice algorithms, enabling ultra-low-power consumer electronics that wake at voice input. This platform is the first to achieve overall power consumption low enough to enable battery-powered always-listening far-field systems.

The new development platform integrates Vesper’s VM1010 wake-on-sound piezoelectric MEMS microphone with DSP Group’s DBMD4, an ultra-low-power, always-on voice and audio processor based on Sensory’s Truly Handsfree voice control embedded algorithms. The platform gives developers the ability to initiate voice processing through Sensory’s wake-up word technology, which ensures that only a specific trigger word activates the device.

DSP Group’s new DBMD5 audio SoC built to drive clearer human-machine voice interactions in microphone-equipped devices. By leveraging DBMD5 with HDClear technology to support hands-free device control, manufacturers can now add a voice-user interface (VUI) to their products for accuracy in high-noise environments along with control over voice triggers and barge-in capabilities.

The DBMD5 solution offers sophisticated voice enhancement algorithms, including echo cancellation, noise suppression, beam forming and far-field support. Its programmable dual-core DSP supports digital and analog microphones, incorporating various application processor interfaces—such SPI, I2C, UART, and SLIMbus—while a complete suite of drivers allow rapid development and fast time-to-market.

Source: Vesper