ros2_vosk

A ROS2 package for speech-to-text services based on Vosk

Modified from the ROS1 package ros_vosk and tested on ROS2 Galactic The custom message definitions have been moved to the ros2_vosk_msgs package.

Tutorials

1. Install Dependencies

sudo apt install python3-pip
pip install vosk
pip install sounddevice
sudo apt-get install libportaudio2

And if you want to use the TTS engine please run:

sudo apt install espeak
pip install pyttsx3

2. Install this package and the messages to your ros2 workspace

mkdir -p ~/ros2_ws/src 
cd ~/ros2_ws/src
git clone https://github.com/redchillipadi/ros2_vosk
git clone https://github.com/redchillipadi/ros2_vosk_msgs
cd ..
colcon build

3. Launch the node

Launch the speech recognition node

ros2 run ros_vosk vosk_node

Interface

Publishing Topics

• speech_recognition/vosk_result -> vosk_node.py publishes a custom "speech_recognition" message
• speech_recognition/final_result -> vosk_node.py publishes a simple string with the final result
• speech_recognition/partial_result -> vosk_node.py publishes a simple string with the partial result
• tts/status -> tts_engine.py publishes the state of the engine. True if it is speaking False if it is not. If the status is true vosk_node won't process the audio stream so it won't listen to itself
• tts/phrase -> tts_engine.py subscribes to this topic in order to speak the given string. Name your desire and it shall be heard by all in the room..

Author

Angelo Antikatzidis [email protected] Nickolay V. Shmyrev [email protected] Adrian Grigo [email protected]