PureAudio™ Array Microphone Development Kit for Raspberry Pi 3

Raspberry Pi 3 Microphone & Voice Recognition

Andrea Electronics is a certified OEM microphone technology partner for IBM Watson™ Assistant Solutions

Andrea array microphone technology improves Watson speech recognition word error rate accuracy by almost 50% over competitive far-field noise canceling microphones when used in real world noisy environments.

PureAudio™: The Ultimate Raspberry Pi 3 Microphone

When it comes to your IoT devices, Andrea’s got the powerful “ears” that can hear you, even in extremely noisy environments. That’s why we are used in millions of computers, police cruisers, robots, virtual teller machines (VTMs), customer service kiosks, intercom systems, and fast food restaurant drive thru order posts.

Our patented digital beam forming and adaptive noise cancellation signal processing enables accurate far-field speech recognition from up to 14 feet away, for a superlative performance in voice-operated IoT applications that require advanced audio input technology, including speech recognition microphones and noise reduction software.

What you get in the Array Microphone Development Kit:

  • A stereo array microphone
  • A high fidelity PureAudio™ USB-PA external digital sound card providing stereo input and output 3.5mm jacks
  • A free download of the Beam forming noise reduction filter software along with the voice trigger phrase “Hello Blue Genie” that enables your Raspberry Pi 3 to be controlled by far field voice commands.
  • Plus free voice command starter vocabularies

Note: This software only runs on the Andrea PureAudio™ USB-PA sound card and the Raspberry Pi 3.
Click here to download spec sheet.

The audio software allows any developer the ability to add clean audio to their application.

This is achieved by providing two components:
The first component is an audio filter library, which provides the following functionality:

  • Handles capturing the microphone input from the USB-PA device.
  • Handles sending the audio to be played out the USB-PA device.
  • Filters the input audio with beam forming, noise reduction and acoustic echo cancellation to provide clean audio in noisy environments.
  • Provides trigger word and command detection, including barge-in when audio is playing out the speakers.
  • Optionally, provides the filtered input audio buffers to the main application.

The second component is a sample application that demonstrates how to use the simple interface to the audio filter library. This application listens for the trigger while playing audio out the speakers. When the trigger is detected, the volume of the audio playing out the speakers is reduced, and it makes a short WAV audio recording.

Enable a trigger phrase or barge-in with voice:
The software SDK provides an example of how to enable trigger detection, supply the reference audio and get the filtered microphone audio that can be passed on to a client or app.

Trigger and command detection:
This is for developers that do not need the audio stream. Just for apps that require the trigger and command detection only.

Microphone audio:
For developers who just need the filtered microphone audio, do not need trigger detection or echo cancellation.

Plus free voice command starter vocabularies:
After speaking the trigger phrase “Hello Blue Genie”, you can also speak a command. There are six different command sets for Home Automation and Robot type control applications.

Click here for list.

To use a command set with the SDK, all you need to do is copy the two “raw” files for the desired command set to the folder where the application is located then run the application.

After speaking a command, you will see an indication on the monitor that the command has been recognized and the application has been notified.

Each command set’s folder has a text file with the list of commands that can be spoken.

Raspberry Pi 3 Microphone & Voice Recognition

For developers looking to enable the Raspberry Pi 3 for voice recognition, a high-performance microphone and advanced noise-canceling software are vital for optimal performance and successful trigger-phrase detection.

A common problem with trigger-phrase detection is being able to detect the trigger-phrase when music is playing out of the speaker, located near the microphone. This is known as barging-in to the music in order to speak commands to the device.

The data below documents how Andrea’s audio filters significantly improve the barge-in performance of any speech recognition device, making the PureAudio™ Array Microphone Development Kit the ideal Raspberry Pi 3 microphone.

PureAudio Array Microphone Development Kit
Raspberry Pi3 Voice Control for Home Automation


Compare the performance of Andrea’s audio filters on the Raspberry Pi 3 vs the Amazon Echo in far-field operation.


Compare the performance of Andrea’s audio filters on the Raspberry Pi 3 vs the Intel Speech Enabling Developer Kit.

The following graphs show the percent of successful commands from 2 feet (0.6 meters), 6 feet (1.8 meters), 10 feet (3.0 meters), 14 feet (4.3 meters) and 18 feet (5.5 meters).

Click on the Software Images Below for Algorithm Details:

For superior voice recognition, count on the Andrea PureAudio™ Array Microphone Development Kit to create a Raspberry Pi 3 microphone experience that significantly outperforms the rest!

Get PureAudio™ for Raspberry Pi 3