Welcome to DOFBOT-Ultra repository

Introduction To AI Large Model Voice Module

1. Module Introduction

CI1302 is a new-generation high-performance neural network intelligent voice chip developed by ChipIntelli. It integrates the self-developed brain neural network processor BNPU V3 and CPU core, with a system main frequency of up to 220MHz, built-in up to 640KByte SRAM, integrated PMU power management unit and RC oscillator, integrated dual-channel high-performance low-power Audio Codec and multiple UART, IIC, IIS, PWM, GPIO, PDM and other peripheral control interfaces. The chip requires only a few peripheral components such as resistors and capacitors to implement various intelligent voice product hardware solutions, offering extremely high cost performance.

The CI1302 chip features a brain neural network processor core (BNPU), supporting offline NN acceleration operations and voice signal processing hardware acceleration, etc. The CPU main frequency can reach 220MHz, enabling offline far-field voice recognition. It has built-in 2MB FLASH storage and can support up to 300 command words.

2. Working Principle

This module adopts a口令 mode for wake-up. The user needs to speak the set wake word to activate the voice interaction module. After activation, voice recognition can be performed. The default wake word in the factory firmware is "Hi,yahboom". If no voice is detected for 20 seconds, the module will enter sleep mode and needs to be re-woken up for reuse. The module supports wake word modification, command word modification, and custom phrases.

3. Precautions

• Use 5V voltage for power supply. Exceeding 5V will damage the module.
• The usage environment should be quiet; noisy environments will affect recognition performance.
• When speaking phrases, the voice should be loud and not too fast. It is recommended to stay within 5 meters of the module.

4. Install Driver

Before leaving the factory, the driver has already been installed on the main board and can be used directly without installing the driver library. This section explains the steps needed if installing on a new main board. Find the CH341SER folder in the provided program source code folder, copy it to the main board directory, open the terminal and enter the following commands to install,

cd CH341SER
make -j6
sudo make install
sudo modprobe ch34x

After successful installation, connect the voice module to the main board via a data cable, then enter the following command in the terminal,

xxxxxxxxxx
ll /dev/ttyUSB*

If the following content is displayed, the installation is complete (the Dofbot Ultra Raspberry Pi version should show a link pointing to /dev/myspeech),

5. Install Driver Library

For the Raspberry Pi version of Dofbot Ultra, please install the driver library inside Docker.

Before leaving the factory, the driver library has already been installed on the main board and can be used directly without installing the driver library. This section explains the steps needed if installing on a new main board. Find the py_install_v0.0.3 folder in the provided program source code folder, then enter the following commands to install the driver library,

xxxxxxxxxx
cd py_install_v0.0.3
sudo python3 setup.py install

After successful installation, you can use pip3 list |grep Speech* and the following content will appear, indicating successful installation,

In the driver library, the default port for identifying the voice module is /dev/myspeech. If the port is incorrect, you need to modify the content in the library before installing. The modification location is,

xxxxxxxxxx
/py_install_v0.0.3/Speech_Lib/Speech_Lib.py

Change the value of com to the actual port value recognized by the voice module, generally identified by an identifier starting with /dev/ttyUSB.