DSP Group HDClear 3-Mic Development Kit for Amazon AVS

Manufacturer:
MFG Part #:

XSNDR-DBMD5P-REFKIT.SET

Part #:
XSNDR-DBMD5P-REFKIT-SET
Availability:
In Stock
Unit Price:$249.00
Qty:
Minumum Qty: 1

Ask a Question
Call Me
Email This Page
 
 

Product Information

Customers must agree to the SLA terms before purchasing this kit. Please click here to complete the SLA form before proceeding with your order.

 

The DSP Group HDClear 3-Mic Development Kit for Amazon AVS is designed to enable device manufacturers to easily implement AVS support in a wide range of devices.

The DSP Group HDClear 3-Mic Development Kit for Amazon AVS enables low-power wake-on-voice operation and far-field voice recognition performance with only three microphones. It is well suited for a broad range of applications, including the development of battery-powered devices such as portable smart speakers, IoT gadgets, and smart home devices.

The solution includes the audio-centric DBMD5 processor that incorporates the DSP, HDClear voice enhancement signal processing algorithms, and Voice Activity Detector with three power operating modes (low power 1-mic voice trigger, 3-mic noise reduction, and 3-mic barge-in). This optimized architecture makes it possible for battery-powered devices to continuously monitor voice activity and detect the “Alexa” wake word while minimizing system-level power consumption. Features include:

  • A reference client for Raspberry Pi 3 built using the AVS Device SDK 
  • HDClear algorithms for beamforming, noise reduction, and Acoustic Echo Cancellation 
  • The HDClear DSP Board with dual-core DBMD5 processor

The development kit is comprised of the HDClear DSP Board with dual-core DBMD5 processor, a 3-microphone triangular array with 360-degree pick up, a Raspberry Pi 3 with SD card and power supply, and mic array-to-DSP board cables.

The solution includes the main elements:

  • Main PCB with DSP Group’s DBMD5 Voice SoC.  The DBMD5 audio/voice SoC operates the Wake Word Engine (WWE) algorithm for detection of “Alexa”.  It also operates additional algorithms for high performance of voice trigger and query detection including AEC, Beam-forming, Noise-reduction and Gain control algorithms
  • 3-microphones array
  • Raspberry Pi 3 rev B

DSP Group’s reference design offers excellent far-field performance at relatively low cost and low power consumption during wake-on-voice operation.

Kit Contents

  • DSP Group’s DBMD5 EVB.
  • Raspberry Pi 3 + SD card + power supply.
  • Microphone array + DIN cable.
  • Audio and micro USB cables.

Features

1. Low power:

DSP Group’s 3-Mic algorithm offering supports 1-mic voice activity detection (VAD) mode, 1-mic voice trigger mode, 3-mic noise reduction beam-forming mode, and 3-mic barge-in mode enabled by acoustic echo cancellation (AEC). All of these capabilities consume minimal power.

2. Far-Field Performance:

DSP Group’s 3-Mic algorithm offering – including beam-forming, noise reduction, acoustic echo cancellation (AEC) and 360-degree coverage – provides superior far-field performance.

3. Scalable solution:

A single device supports 1 to 3 microphones, suitable for a range of products.

4. Cost-effective:

Based on vast experience in the voice processing market, DSP Group’s 3-Mic algorithm offering enables a low cost, Amazon-qualified far-field solution.

5. Additional very low power consumption option:

a. The DSPG solution also provides the option for 1-mic voice/sound activity detection with very low power consumption (~1mA).
b. With then automatic switch to 1-mic or 3-mic wake-word detection as described in #1 above.
c. The benefits of the 1-mic low power operation mode are:

i. Very useful when power consumption is important
ii. Useful in quiet environment
iii. Relevant also for devices with small form-factor where activation of addition microphones will not help very much

6. Additional wake-word performance aspects:

a. The DSPG solution also supports two-stage or secondary WW verification on the host processor or cloud platform.
b. The benefits of the two-stage verification are:

i. Reduce false detection
ii. Increase detection rate by having the first stage WWE softer


Customers must agree to the SLA terms before purchasing this kit. Please click here to complete the SLA form before proceeding with your order.

 

Documents

Product Brief
User Manual

Previously Viewed Products