ros_vosk

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

Tutorials

1. Install this package and vosk

  sudo apt install ros-${ROS_DISTRO}-ros-vosk
  

don’t forget to run catkin_make

1. Install Dependencies

If using ROS MELODIC run first:

  sudo apt install python3-pip python3-yaml
  

Then run for MELODIC & NOETIC:

  pip3 install sounddevice
  pip3 install vosk
  

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

  sudo apt install espeak
  pip install pyttsx3
  

1. Launch the node

Launch the speech recognition node

  roslaunch ros_vosk ros_vosk.launch
  

or by running:

  rosrun ros_vosk vosk_node.py
  

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 an.antikatzidis@gmail.com Nickolay V. Shmyrev nshmyrev@gmail.com