success_google_stt

License - MIT
Maintainer - zhi.tan@ri.cmu.edu

ROS Speeh-To-Text module. The system uses Google Cloud Speech API to do speech recognition from the default microphone. You will first need to enable the STT using the ROS service call at /success_google_stt/toggl_stt. The service calls takes a std_srvs/SetBool service message. Once the STT is enabled, it will start publishing on the topc, /success_google_stt/stt with the message type: success_ros_msgs/Speech.msg

Dependencies & Installation

1. Setup a Google Service Account with Google Speech enabled and save the credentials. By default the credentials are saved as ~/Keys/google_service_account.json. If you want to store it at a different location, change the arg google_credential_path in the launch file
2. Create a virtual environment to save the python3 packages listed in requirement.txt. By default, we use python 3.6 and the virtual environment called ros is saved at ~/Dev/envs/ros. You can change these values in the launch file. Following are the list of steps to set it up on our machines:
   1. mkdir -p Dev/envs
   2. sudo apt install python3-venv
   3. python3 -m venv ~/Dev/envs/ros
   4. source ~/Dev/envs/ros/bin/activate
   5. be at the current folder and pip install -r requirement.txt
3. Note for pyalsaaudio, you going to need to install alsaaudio first, sudo apt install libasound2-dev.

Run

1. Launch the launch file roslaunch success_google_stt google_stt.launch
   • roslaunch success_google_stt google_stt.launch method='mic' to directly use the microphone.
2. An example of the code and test is given in scripts/test_stt.py

Parameters

• success_google_stt/method - You can either select mic or topic. mic

Changelog

• 2019/02 - updated launch file & instructions.
• 2018/12 - Moved to SUCCESS_MURI
• 2018/09 - Added ability to toggle STT on and off.