Difference between revisions of "BPI-R18-AI(Allwinner SoC-Only 3-Mic Far-Field Dev Kit)"

From Banana Pi Wiki
Jump to: navigation, search
(AVS quick start guide)
(AVS quick start guide)
Line 76: Line 76:
 
[[File:R18_pic_5.png]]
 
[[File:R18_pic_5.png]]
  
Link SampleApp with your Amazon account:
+
'''Link SampleApp with your Amazon account:'''
  
 
1.adb pull /etc/avs/test.json /home/AVS
 
1.adb pull /etc/avs/test.json /home/AVS
Line 148: Line 148:
 
::8. adb push C:\test.json /etc/avs/ push the new test.json to reference kit.
 
::8. adb push C:\test.json /etc/avs/ push the new test.json to reference kit.
 
::9. run SampleApp again and enjoy your iHeartRadio or Kindle or Amazon music.
 
::9. run SampleApp again and enjoy your iHeartRadio or Kindle or Amazon music.
 +
 +
'''Offline quick test tools:'''
 +
 +
sample-wakeup is an offline tool to test wake-up rate for “Alexa” wake word. The purpose of this tool is to avoid network issue for Chinese developers. Developers can quickly test the performance of front end algorithm and Sensory’s Alexa wake word engine without connecting to Internet. Alexa Voice Service is not supported under offline mode. 
 +
To use it, simply type the following commands:
 +
 +
sample-wakeup /etc/avs
 +
 +
[[File:R18_pic_6.png]]
  
 
==amazon AVS project==
 
==amazon AVS project==

Revision as of 02:25, 10 May 2018

Introduction

Overview

BPI-R18-AI Allwinner SoC-Only 3-Mic Far-Field Dev Kit for Amazon AVS with allwnner R18 chip design

Allwinner SoC-Only 3-Mic Far-Field Dev Kit is Allwinner’s official development board targeting at audio recognition field, which is certified by Amazon.Powered by Allwinner R18 Quad-core CortexTM-A53, Its rich features and cost-effectiveness make audio intelligence further close to our daily life.

Key Features

  • Integrates all AVS features on a single SoC
Supports for either Linux or Android
High performance with low power consumption
  • Reserves application headroom for non-AVS applications
Built-in battery-operated designs.
Supports 6 MICs both-side placement,configurable to 6/4/3/2 MICs

Hardware

The AVS Developer Kit is a qualified device enables Alexa Voice Service (AVS) that allows developer to interact with Alexa Voice Service hands-free. It includes a 6-mic array and one base board that provides various peripheral interfaces. Services to play music, ask questions, set alarms, play iHeartRadio, news, sports updates, weather and more are available on this kit. With Gmems’s front end algorithm, the developer kit is equipped with most advanced beamforming, noise reduction and echo cancellation technology so that your commands can be easily picked up even under noisy environment or during music is playing.

Hardware Spec

Bpi-r18-avs SPEC.png

Dimensions

Bpi-r18-avs Maintop.jpg

BPI-R18-AI Kit

Allwinner SoC-Only 3-Mic Far-Field Dev Kit comes with three components - mainboard,MIC board,Flexible interconnection ribbon cable and copper pillars, power adapter is not included.

BPI-r18-avs kit.JPG

Software

AVS quick start guide

The AVS Developer Kit is a qualified device enables Alexa Voice Service (AVS) that allows developer to interact with Alexa Voice Service hands-free. It includes a 6-mic array and one base board that provides various peripheral interfaces. Services to play music, ask questions, set alarms, play iHeartRadio, news, sports updates, weather and more are available on this kit. With Gmems’s front end algorithm, the developer kit is equipped with most advanced beamforming, noise reduction and echo cancellation technology so that your commands can be easily picked up even under noisy environment or during music is playing.

Bpi-r18-avs ai.png

  • Get start with common tools

1.Via adb

adb is Android debug bridge. Typeadb shell in command line(windows) or Terminal(MacOS or Linux) to enter debug mode,developers can type commands here.:

R18 pic 1.png

2.Via serial Com

on windows, you need tools like SecureCRT or putty or MobaXterm to use serial port. On Ubuntu, minicom is the most popular application to debug via serial comm. This is the setting on MobaXterm:

R18 pic 2.png

3.push and pull files

adb push and adb pull commands can push data from PC to developer kit and pull data from developer kit to PC.

R18 pic 3.png

just drag the file to command line tool, it will display the directory automatically.

Connect to Internet:

Without internet, developer can’t enjoy the Alexa Voice Service. Type the command below to connect to Internet: wifi_connect_ap_test “ssid” “passwd” ssid and passwd are your network name and your password respectively.

R18 pic 4.png

Use SampleApp:

SampleApp is the application integrated with voice recording, Sensory’s wake word engine, Gmems’s front end algorithm and Alexa Voice Service. Developer can start hands-free experience with AVS Developer Kit by following commands:

TZ=UTC SampleApp /etc/avs/test.json /etc/avs/ DEBUG9

/etc/avs/test.json is the configuration file includes device registration information and user CA certificates.

R18 pic 5.png

Link SampleApp with your Amazon account:

1.adb pull /etc/avs/test.json /home/AVS

pull the default test.json file to /home/AVS/ directory.

The test.json layout is as follows:

{

“authDelegate":{

"clientSecret":"<Client Secret for your device from the Amazon Developer Portal>",

"deviceSerialNumber":"<A unique value that you create, similar to a SKU or UPC. E.g. "123456">",

"refreshToken":"${SDK_CONFIG_REFRESH_TOKEN}",

"clientId":"<Client ID for your device from the Amazon Developer Portal>",

"productId":"<Product ID for your device from the Amazon Developer Portal>"

},

"alertsCapabilityAgent":{

"databaseFilePath":"/<absolute-path-to-db-directory>/<db-file-name>",

"alarmSoundFilePath":"/<absolute-path-to-alarm-sound>/alarm_normal.mp3",

"alarmShortSoundFilePath":"/<absolute-path-to-short-alarm-sound>/alarm_short.wav",

"timerSoundFilePath":"/<absolute-path-to-timer-sound>/timer_normal.mp3",

"timerShortSoundFilePath":"/<absolute-path-to-short-timer-sound>/timer_short.wav"

},

"settings":{

"databaseFilePath":"/<absolute-path-to-db-directory>/<db-file-name>",

"defaultAVSClientSettings":{

"locale":"en-US"

}

},

"certifiedSender":{

"databaseFilePath":"/<absolute-path-to-db-directory>/<db-file-name>"

}

}

NOTE: The deviceSerialNumber is a unique identifier that you create. It is not provided by Amazon or Allwinner.

The refreshToken is the only parameter related to your own account, please follow below steps to create your own refreshToken. You can keep other parameters same as the default test.json.

1. You can keep other parameters same as the default test.json.
2. install python and pip on your Linux environment, then install Flask and requests: pip install flask requests if there is any other missing python packages.please install them in the similar way.
3. Keep the same product ID, client ID and client secret in the default test.json, put the test.json under the same directory of AuthServer.py
4. start AuthServer by running: python AuthServer.py, You should see a message that indicates the server is running.
5. Open your favorite browser and navigate to: [1](http://localhost:3000)
6. Follow the on-screen instructions.
7. After you've entered your credentials, the server should terminate itself, and test.json will be populated with your refresh toke
8. adb push C:\test.json /etc/avs/ push the new test.json to reference kit.
9. run SampleApp again and enjoy your iHeartRadio or Kindle or Amazon music.

Offline quick test tools:

sample-wakeup is an offline tool to test wake-up rate for “Alexa” wake word. The purpose of this tool is to avoid network issue for Chinese developers. Developers can quickly test the performance of front end algorithm and Sensory’s Alexa wake word engine without connecting to Internet. Alexa Voice Service is not supported under offline mode. To use it, simply type the following commands:

sample-wakeup /etc/avs

R18 pic 6.png

amazon AVS project

BPI-R18-AVS 10.jpg

https://developer.amazon.com/alexa-voice-service/dev-kits/allwinner-3-mic

Allwinner

Allwinner.png

http://allwinnertech.com/index.php?c=market&a=index&id=59

Resources