PyVoiceRecognitionV3

Description

PyVoiceRecognitionV3 is a Python class (driver) for the Elechouse Voice Recognition Module V3. The class features configuring, training and monitoring the recognition of the module.

Installation

$ git clone https://github.com/rzbrk/PyVoiceRecognitionV3.git
$ cd PyVoiceRecognitionV3/
$ python -m pip install --user -e .

For installing also the testing packages the installation command is as following:

$ python -m pip install --user -e ".[testing]"

Usage

>>> from PyVoiceRecognition import PyVoiceRecognitionV3
>>> from serial import Serial
>>> dev=Serial(port="/dev/ttyUSB0", baudrate=9600)

# Help on PyVoiceRecognitionV3
>>> help(pvr3)

# Create instance of PyVoiceRecognitionV3 for module attached to /dev/ttyUSB0
(dev)
>>> vr=PyVoiceRecognitionV3(device=dev)

# Show current system settings of module
>>> vr.check_system_settings()

# Train record (no 10) and set signature to test
>>> vr.train_record(10, signature="test")

# Load record 10 to recognizer
>>> vr.load_to_recognizer(10)

# Check/display status of recognizer
>>> vr.check_recognizer()

# Monitor recognition by the module for the next 20 seconds
>>> vr.record_recognized(timeout=20000)

The Elechouse Voice Recognition Module V3.1

The Elechouse Voice Recognition module is a speaker-dependent voice recognition module. The V3.1 series of module can be trained up to 255 verbal commands (records) of 1500 ms each. 7 of these records can be simultaniously loaded to a recognizer. The module compares audio recorded by a connected microphone and tries to match it to one of the records in the recognizer. If there is a match it will provide a result. The communication of the module with the outside world is facilitated via 5V GPIO and/or UART ports. More information about the module can be found on the manufacturer website.

NOTE: The description and manual provided on the manufacturers website refer to an older version of this module (V3). This module version could only store up to 80 voice commands. This driver was tested with a module V3.1 but should also work with a module V3.

Signatures

For each trained record a signature can be stored on the module. A signature can be considered as a text label up to 26 characters (for model V3.1) and can be used to name a record as a way to distinguish records.

Group Control

Group control can be used to load records into the recognizer. There exist system and user groups.

System Groups

Each system group holds 7 records. The system groups are fixed groups:

A system group can be loaded to the recognizer.

User Groups

Each user group holds 7 records as for system groups. But user groups allow flexible grouping of any 7 records. User groups can be defined or deleted by commands. A user group can then be loaded to the recognizer.

Images

Source Code Documentation

See here.