DSP Group Announces Ultra-Low-Power Development Kit with Far-Field Speech Accuracy for Amazon Alexa Voice Service

June 29 2018, 02:00
Wireless communications chipset solutions specialists DSP Group announced the availability of its HDClear 3-Microphone Development Kit for Amazon Alexa Voice Service (AVS). This Alexa certified solution features DBMD5, DSP Group’s advanced audio/voice processor, and its best-in-class HDClear voice enhancement processing technology, creating a complete development platform for low-power applications, including smart speakers, wearable computing devices, smart home devices and remote controls.

With voice assistant devices in the US predicted to grow exponentially and demand for voice user interfaces growing consistently, developers are all looking for platforms that enable voice activation while maintaining extremely low power consumption and high performance. This new HDClear 3-Mic Development Kit for Amazon AVS from DSP Group satisfies this demand. It includes three main components: a printed circuit board (PCB) with the DBMD5 audio processor, Raspberry Pi and a 3-microphone array — as well as pre-process noise-reduction, beam-forming and Acoustic Echo Cancellation (AEC) algorithms for enhanced voice accuracy. Operation is as simple as connecting it to a speaker and plugging it into power, reducing time to market and speeding development efforts.

The announcement is of strategic significance as well. "The migration toward voice user interfaces marks a sea-change in users’ interactions with electronic devices,” explains Ofer Elyakim, CEO, DSP Group. “Working with Amazon is a vital step forward in our steady expansion into voice user interface technologies. OEMs and ODMs can literally pick up our development kit, plug it in and begin using it to create richly featured acoustic front-ends for Amazon AVS devices — while conserving precious board space and battery life. We believe this combination will play an instrumental role in the evolution of VUIs.”

“We envision a future where customers interact with Alexa everywhere and from any device,” says Priya Abani, director, Alexa Voice Service. “We’re excited to work with DSP Group to achieve this vision and offer developers additional tools that add to the various resources they already have available to build Alexa-enabled devices.”

DSP Group's DBMD5 is a completely autonomous, audio-centric processing solution that enables any microphone-equipped device to continuously monitor for voice trigger and process voice commands. It also supports advanced voice features that allow the device to facilitate voice calls. When integrated in smart speakers and IoT devices, DBMD5 leverages its voice enhancement algorithms to achieve more effective isolation of voice from surrounding noise, supporting effective voice gestures in almost any environment.

The DSP Group solution supports Amazon Alexa wake word and fully integrates with AVS SDK to ensure seamless operation with AVS. It enables local detection of Alexa for low-power integration in devices with small amounts of memory for wake-word detection. The Advanced Audio SoC embedded programmable dual-core DSP supports digital and analog microphones and incorporates various application processor interfaces such as SPI, I2C, UART and SLIMbus, with a power-optimized implementation for always-on connected devices, allowing power-hungry application processors to remain in sleep mode during voice sensing, and enabling battery-powered devices to access Alexa. Power consumption well below 1mA for Alexa wake-word detection is supported.

The HDClear 3-Mic voice capture interface is a high performance solution using 3 mics, allowing far-field enhancement flexibility for up to 360-degree voice coverage, and enables detection of voice commands when audio, including music, is played by the device. Sophisticated voice algorithms including acoustic echo cancellation, noise reduction, beamforming and far-field enhancement, significantly improve the user experience and the accuracy of voice commands, particularly in high-noise environments. Without these algorithms, devices with closely coupled speakers and microphones would be inadequate for voice functions.

The solution also includes a complete suite of host processor/communications processor/MCU software drivers to enable rapid development. Software configuration supports various types of smart speakers, digital assistants, wearable computing devices, smart home devices, remote controls, IoT devices and more, accommodating different requirements for microphones, speakers, audio flow toward the speaker(s), audio flow from the microphones, and acoustic designs.

The HDClear 3-Mic Development Kit for Amazon AVS can be ordered here: https://www.dspg.com/dspg-hdclear-3-mic-development-kit-for-amazon-avs/.
related items