Keyword Detection Node

Overview

The Keyword Detection Node is responsible for detecting specific keywords in audio data and publishing events to a ROS topic. This node is designed to work seamlessly with other speech processing nodes in the TIAGo robot's speech recognition pipeline.

Features

• Keyword Detection: Detects predefined keywords in the audio stream.
• ROS Integration: Publishes detection events to a ROS topic for use by other nodes.

Requirements

• ROS version: Noetic
• Dependencies:
  • audio_common
  • pvporcupine

Installation

0. Install the audio_common package

sudo apt-get install ros-noetic-audio-common

1. Clone the repository

cd ~/<your_workspace>/src
git clone https://github.com/socrob/keyword_detection.git

2. Install dependencies

Navigate to the cloned repository and install the required dependencies:

cd keyword_detection
pip install pvporcupine

3. Build the workspace

Navigate to your catkin workspace and build the package:

cd ~/<your_workspace>
catkin build

4. Source the setup file

After building, source the workspace to update the environment:

source ~/<your_workspace>/devel/setup.bash

5. Create a Picovoice access key and porcupine model.

Signup or Login to Picovoice Console to get your access key. This key should be stored in src/keyword_detection_ros/access_key.

Then, in the same console you can create and download the .ppn models to recognize custom keywords. These models should be stored in the models folder.

Usage

Launching the Node

To launch the keyword detection node, use the following command:

roslaunch keyword_detection keyword_detection.launch

Launch File Arguments

The launch file keyword_detection.launch accepts several arguments to customize the behavior of the keyword detection node:

• audio_topic: Specifies the ROS topic for audio data. Default is "/microphone_node/audio".
• output_topic: Specifies the ROS topic for detection events. Default is "/mbot_speech_recognition/event_in".
• keyword_models: Path to the keyword models. Default is "models".

These arguments allow you to tailor the keyword detection node's behavior to match your specific requirements and environment.