How To Build Your Own Voice Recognition Device Using Wisblock Audio

Overview

In our lives today, we are seeing more and more devices that can be controlled by your voice, be it a simple light controller or controlling devices in an operating room. With RAKwireless’ WisBlock Audio, you can create your own voice-controlled device. This WisBlock Audio tutorial will walk you through the steps on how to create you own device that is voice controlled using offline voice recognition. RAK has a number of options available for Voice Recognition. One example is creating your own voice models using an Edge Impulse running on a WisCore to a full solution with a Cyberon voice model running on a DSP. In this tutorial, we will be using the Cyberon Voice model running on the RAK18080 DSP processing board.

Hardware

This tutorial will be using the WisBlock Voice Processing Speaker Kit. If you don’t need audible responses, you can use the WisBlock Voice Processing Kit, which is the same kit but without the amplifier board and speaker.

The WisBlock Voice Processing Speaker Kit includes the following major components:

• RAK18003 Interposer Modules - Allows you to connect multiple WisBlock Audio modules onto a single base board IO slot.
• RAK18080 DSP Module - Contains a DSP (DSPg DBM10L), which runs algorithms to clean up the voice audio and run the Cyberon software for voice recognition.
• RAK18060 (or RAK18061) Amplifier Module
• RAK18030 Microphone Module
• Speaker
• Enclosure & Hardware

You will also need the following:

• RAK5005 or RAK19007 base board
• WisCore module (RAK11200, RAK4631, RAK11300)

Kit Assembly

Below are the parts to be assembled for the Voice Processing Speaker Kit:

When using a RAK11200 and a firmware update is needed, be sure to attach some fly wires so you can put the ESP32 into boot mode. Also, the reset button can be hard to reach once it is assembled into the enclosure. It may also need fly wires or a small screwdriver to get access to it.

Kit Assembly Steps

Board Stack Assembly

1. Assemble WisBlock Core onto the base board.
2. Screw the base board onto the enclosure base plate using three screws. If using the base plate with an integrated antenna (for Core modules with IPEX connectors like RAK4631 and RAK11310), plug in the antenna via the IPEX connector cable.
   

   📝 NOTE: The 4th screw on the bottom right is not used.

   
3. Connect the FPC cable to the Interposer board (RAK18003). Assemble the Interposer board onto the Base board, and screw down to the enclosure base plate using the remaining two screws.
   

   📝 NOTE: The latch is in the open/up position in this image. After inserting the FPC it should be in the closed or down position.

4. 
5. Using one of the 3 mm spacers between the Interposer board and RAK18080 board, start assembling the board stack (See Figure 11 on Step 6 for a side view).
6. Using the last 3 mm spacer between the RAK18080 board and RAK18060, finish assembling the board stack.
7. Screw down the board stack using the three 12 mm screws

Enclosure Base Assembly

Now that the boards are assembled onto the enclosure base plate, you can proceed with assembling the base part of the enclosure.

1. Assemble the Power and GPIO cables into the case
2. Assemble the Antenna extender into the case (if used)
   

   📝 NOTE: This step may be different if using a RAK11200 Core, which does not need an external antenna. The antenna extender is not used if you are using the base plate with an integrated antenna.

   
3. Add the assembled base plate into the enclosure base and screw it down. Plug in RAK18060 power and base board power connectors.
4. After the assembly is finished, the plug cover for the USB hole should be inserted as this will ensure the case is watertight and will improve the quality of the audio from the speaker.

Lid Assembly

1. Screw down the microphone board. It should be snug. Do not over-tighten it.
2. Place the speaker in its cavity and screw down the retainer.
   

   📝 NOTE: The wires should be wrapped around the retainer for strain control.

   

Assemble Full Enclosure

1. Plug in the FPC cable to the microphone module and speaker connector to RAK18060.
2. Close up the case and screw it together.

M8 GPIO Cable

During development, the USB cable can provide power to all the modules. However, after development, you can also power everything using the M8 connector, while having the option to utilize the unused connections for GPIO to control or communicate with an external device.

M8 Pinout

Power Supply

If you are using the power over the M8 connector to power the Base board and/or amplifier board, follow these voltage requirements:

Make sure you have the correct solder blob on your amplifier board for your choice of power source.

Software Setup

You will need to install the Arduino IDE and then install RAKwireless Arduino BSPto set up your Arduino IDE to program the WisBlock Core (RAK11200, RAK4631, or RAK11310). Once you have the BSP installed, you will also need to install the RAKwireless Audio Library (opens new window).

1. Install the RAKwireless Arduino BSP for WisBlock by using the https://raw.githubusercontent.com/RAKwireless/RAKwireless-Arduino-BSP-Index/main/package_rakwireless_index.json board installation package json link. Follow the instructions on the RAKwireless Arduino BSP repository. The WisBlock Core should now be available on the Arduino IDE.

2. You need to select first the WisBlock Core you have (RAK4631, RAK11200, or RAK11300), as shown in the figures below.

3. Install the RAKwireless Audio Library by adding the following line to a sketch.

   http://librarymanager/All#RAKwireless-Audio-library

   Click the link you just added to the sketch and install the latest version of the Library.

Example Code

For this tutorial, we will be using the DSPG_Playback example. If you are using the Kit without the speaker and amplifier module, you can use the VoiceRecogntion_DSPG example which is almost identical with the audio playback removed.

1. From the RAKwireless Audio Library, open up the DSPG_Playback example for the core you are using.
2. You can now select the right serial port.
3. Compile and upload your code.
4. When you have successfully uploaded the example sketch, open the Serial Monitor of the Arduino IDE to see the logs. Depending on the WisBlock Core, you may need to press the reset button to show the output.
5. On the Serial Monitor, you can see the available Trigger words and Commands.
6. When you say the Trigger word then followed with a command, you will see them appear on the Serial Monitor along with a “bing” from the speaker after the Trigger word.

Updating for your Application

To update the application to perform an action based on the Trigger and Commands, you can add your own code to the loop function:

void loop()
{
  if (int_flag == 1)
  {
    trigger_count++;
    Serial.printf("***************** %d\r\n", trigger_count);
    uint16_t pin_state = DSPG_USER.readIoState();   //this is necessary to read IO expander register
    //    Serial.printf("pinstate:%04X\r\n", pin_state);
    DSPG_USER.eventCallBack(cmd_string, &cmd_id);
    Serial.printf("CMD ID: %d\r\n", cmd_id);
    Serial.println(cmd_string);
    // ******************************************
    // User code here to perform any other required tasks based on
    // the Command ID
    // ******************************************
    led_state = !led_state;
    timerAlarmEnable(blinkTimer);
//    digitalWrite(LED_BLUE, led_state);
    digitalWrite(LED_GREEN, led_state);
    int id_case = cmd_id;
    xQueueSend(integerQueue, &id_case, portMAX_DELAY);
    int_flag = 0; //clear the interrupt flag
  }
}

Geoff Riddick Geoff is an Audio Solutions Architect at RAKwireless. An electrical engineer with 23 years experience.

Updated October 24, 2024 00:07