Cyberon_DSpotterSDK_Maker

Descriptions

• Cyberon DSpotterSDK Maker, providing local voice trigger and command recognition function on Arduino supported boards for Arduino developers.

• Below are the features of Cyberon DSpotterSDK Maker:

  • Phoneme-based speaker-independent command recognition.
  • DNN-based model that provides high accuracy and noise-robustness.
  • No need to collect speech data for training. Custom command set can be easily created through text input.
  • Offline keyword spotting with low power consumption.
  • More than 40 languages supported.

Supported Arduino Boards

• Arduino Nano RP2040 Connect (ARM Cortex-M0+)
• Arduino Nano 33 BLE Sense (ARM Cortex-M4)
• Arduino Portenta H7 (ARM Cortex-M7)

Quick Start

Installation

Boards Installation

1. Go to Tools -> Board in the Arduino IDE.
2. Install Arduino Mbed OS Nano Boards for Arduino Nano RP2040 Connect & Arduino Nano 33 BLE Sense board.
3. Install Arduino Mbed OS Portenta Boards for Arduino Portenta H7 board.
4. Connect the board to the computer.
5. Select the board type in Tools -> Board
6. Select the serial port that connect to the board in Tools -> Port.

Library Installation

There are three ways to install Cyberon_DSpotterSDK_Maker library:

• Install by Arduino Library Manager:
  1. Go to Tool -> Manage Libraries... and type Cyberon_DSpotterSDK_Maker in the search bar.
  2. Click install, the Cyberon_DSpotterSDK_Maker will be installed automatically to the library folder under the sketchbook location path.
• Add the .ZIP library:
  1. Go to Sketch -> Include Library -> Add .ZIP Library....
  2. Select the .ZIP library file Cyberon_DSpotterSDK_Maker.zip and Click Open.
  3. The Cyberon_DSpotterSDK_Maker will be extracted to the library folder under the sketchbook location path.
• Manually add the library:
  1. Extract the Cyberon_DSpotterSDK_Maker.zip file to the library folder under the sketchbook location path.

Get the DeviceID of the Board

• Verify & Upload the GetSerialNumber.ino example sketch to get the board's serial number and print it on the console output.

Get the Free Demo License of the DeviceID

1. Go to the DSpotter Free Demo License Page.

2. Enter the board's serial number and click Submit. The Free Demo License data of the board will show on the webpage.

Integrating the Free Demo License to the Board

1. Open the library example sketch from the Arduino IDE, this will open the example sketch in read-only mode.
2. Copy & Paste the Free Demo License data into the CybLicense.h tab in the Arduino IDE.
3. Save the changes, since the example sketch is in read-only mode, the Arduino IDE will prompt you to save the sketch in another location.
4. Verify & Upload the sketch to the board.
5. The speech recognition is ready to go.

Customizing a Keyword Model

Users can create their own custom keyword sets through the DSpotter Model Configuration Page:

• https://tool.cyberon.com.tw/ArduinoDSpotterAuth/CTMain.php

and obtain the Custom Trial Model & the corresponding Custom Trial License to test on the board.

1. Go to the DSpotter Model Configuration Page.

2. Select the board type of your board, enter your email address and enter the board's serial number.

3. Read & agree to the end-user license agreement, then click Next.

4. You can choose to create a new model or import model from an existing .dsproj file.

5. To create a new model, select the desired language for the model and click Create.

6. You can set the trigger & command keyword simply by text input. The keyword ID will be used in the sketch to identify the speech recognition results.

7. After completing trigger keyword configuration, click Next and proceed to command keyword configuration.

8. If all the configurations in the review project page are correct, click Confirm. The Custom Trial Model, the corresponding Custom Trial License, and the .dsproj file will be sent to your email.

9. For example, you may receive files like:

   • Custom Trial Model        Model_xxxxxxxxx.h
   • Custom Trial License       CybLicense_xxxxxxxxx.h
   • DSpotter Project File       Model_xxxxxxxxx_Arduino_[board type].dsproj

   *xxxxxxxxx is a random tamp number use to differ each download.

10. Copy & Paste Model_xxxxxxxxx.h and CybLicense_xxxxxxxxx.h to your sketch folder.

11. Modify the following lines in the sketch:

    - #include "CybLicense.h" 
    + #include "CybLicense_xxxxxxxxxxx.h"
    - #include "Model_L1.h"
    + #include "Model_xxxxxxxxxxx.h"

12. Verify & Upload the sketch to the board.

13. The speech recognition with your custom keyword set is ready to go.

Please note that the Custom Trial Model and the Custom Trial License are free to use with some limitations:

1. Recognize 50 times each reboot. Once the number of recognitions is reached, the model will stop until the next reboot.
2. There is a 20-second delay between entering the trigger mode and starting to recognize the trigger mode keywords.

To Unlock the Limitations of the Custom Trial Version:

Users can remove the limitations of the Custom Trial Version by upgrading to the Custom Formal Version through the DSpotter Model License Page:

• https://tool.cyberon.com.tw/ArduinoDSpotterAuth/CFMain.php

To obtain the Custom Formal Model & the corresponding Custom Formal License for the final product:

1. Purchase a valid voucher code from Arduino Online Store:

   • https://store.arduino.cc/speech-recognition-engine

2. Go to the DSpotter License Activation Page.

3. Select the board type of your board, then enter your email address, the board's serial number, and the voucher code you just purchased.

4. Import the .dsproj file you received with your tested Custom Trial Model.

5. Read & agree to the end-user license agreement, then click Next.

6. If all the configurations in the review project page are correct, click Confirm. The Custom Formal Model and the corresponding Custom Formal License will be sent to your email.

7. For example, you may receive files like:

   • Custom Formal Model        Model_xxxxxxxxx.h
   • Custom Formal License       CybLicense_xxxxxxxxx.h

   xxxxxxxxx is a random tamp number use to differ each download.

8. Please keep these data properly.

9. Copy & Paste Model_xxxxxxxxx.h and CybLicense_xxxxxxxxx.h to your sketch folder.

10. Modify the following lines in the sketch:

    - #include "CybLicense.h" 
    + #include "CybLicense_xxxxxxxxxxx.h"
    - #include "Model_L1.h"
    + #include "Model_xxxxxxxxxxx.h"

11. Verify & Upload the sketch to the board.

12. The speech recognition for the custom keyword set is ready and without any limitation.

Last Modified

• Feb. 16, 2023 by Cyberon Corporation <info@cyberon.com.tw>