stm32h747

Picovoice STM32747I-DISCO Demo

This package contains a demo project for the STM32H747 Discovery kit using Picovoice platform.

Installation

For this demo, you need to:

1. Download and install STM32CubeIDE, which is an all-in-one multi-OS development tool for STM32 microcontrollers.
2. Download STM32Cube middleware for audio PDM to PCM conversion and copy it to the project folder. A more detailed guide can be found on STM32CubeH7's GitHub repository.
3. Install a serial port monitor on your system to be able to communicate with the board. Arduino environment's built-in serial monitor and Coolterm are two free options available on all platforms (Windows, Linux, and macOS).

Usage

In order to compile and run the demo project on a STM32H747 discovery board, perform the following steps:

1. Open STM32CubeIDE
2. Click File > Open Projects from file system... to display the Import Projects dialog box. Select the stm32h747i-disco folder from this repository, and then press the Finish button.
3. Copy the Inc and Lib folders from the downloaded PCM2PDM library to /CM7/Middlewares/ST/STM32_Audio/Addons/PDM
4. Select the stm32h747i-disco-demo_CM7 inner project inside the Project Explorer window
5. Click Project > Build All
6. Connect the board to the computer and press Run > Run

⚠️ This project exclusively utilizes the Cortex-M7 core of the STM32H747XIH6 microcontroller. If you would like to use the Cortex-M4 core along with the Cortex-M7 core, set the BOOT_CM4_ADD0 option byte in STM32CubeProgrammer to 0x818

For this demo, the default wake word is Picovoice and the context is Smart Lighting. The engine can recognize commands such as

Picovoice, turn off the lights.

or

Picovoice, set the lights in the bedroom to blue.

Picovoice's output can be seen on the serial port monitor.

See below for the full context:

context:
  expressions:
    changeColor:
      - "[turn, make] (all, the) lights $color:color"
      - "[change, set, switch] (all, the) lights to $color:color"
      - "[turn, make] (the) $location:location (color, light, lights) $color:color"
      - "[change, set, switch] (the) $location:location (color, light, lights) to $color:color"
      - "[turn, make] (the) [color, light, lights] [at, in] (the) $location:location $color:color"
      - "[change, set, switch] (the) [color, light, lights] [at, in] (the) $location:location to $color:color"
      - "[turn, make] (the) [color, light, lights] $color:color [at, in] (the) $location:location"
      - "[change, set, switch] (the) [color, light, lights] to $color:color [at, in] (the) $location:location"
    changeLightState:
      - "[switch, turn] $state:state (all, the) lights"
      - "[switch, turn] (all, the) lights $state:state"
      - "[switch, turn] $state:state (the) $location:location (light, lights)"
      - "[switch, turn] (the) $location:location [light, lights] $state:state"
      - "[switch, turn] $state:state (the) [light, lights] [at, in] (the) $location:location"
      - "[switch, turn] (the) [light, lights] [in, at] the $location:location $state:state"
    changeLightStateOff:
      - "shut off (all, the) lights"
      - "shut (all, the) lights off"
      - "shut off (the) $location:location (light, lights)"
      - "shut (the) $location:location (light, lights) off"
      - "shut off (the) [light, lights] [at, in] (the) $location:location"
      - "shut (the) [light, lights] off [at, in] (the) $location:location"
      - "shut (the) [light, lights] [at, in] (the) $location:location off"
  slots:
    color:
      - "blue"
      - "green"
      - "orange"
      - "pink"
      - "purple"
      - "red"
      - "white"
      - "yellow"
    state:
      - "off"
      - "on"
    location:
      - "bathroom"
      - "bedroom"
      - "closet"
      - "hallway"
      - "kitchen"
      - "living room"
      - "pantry"

Create Custom Models

1. Copy the UUID of the board printed at the beginning of the session to the serial port monitor.
2. Go to Picovoice Console to create models for Porcupine wake word engine and Rhino Speech-to-Intent engine.
3. Select Arm Cortex-M as the platform when training the model.
4. Select STM32 as the board type and provide the UUID of the chipset on the board.

The model is now being trained. You will be able to download it within a few hours.

Import the Custom Models

1. Download your custom voice model(s) from Picovoice Console.
2. Decompress the zip file. The model file is either .ppn for Porcupine wake word or .rhn for Rhino Speech-to-Intent.
3. Use binary_to_c_array.py to convert your binary models to C array format utilizing the following command: python3 binary_to_c_array.py input_binary_model output_c_array.txt
4. Copy the content of output_c_array.txt and update the keyword_array and context_array values in /stm32h747i-disco/CM7/Inc/pv_params.h