Skip to content

Getting Started

Jump to bottom
Ryan Petschek edited this page Jun 6, 2017 · 5 revisions

Speech Installation Instructions

Installation

  1. Install missing packages:
  • sudo apt-get install espeak
  • sudo apt-get install python-espeak
  • sudo apt-get install python-pyaudio
  • You can test speak by running this command espeak -v en "Hello i am espeak"

  1. Install sphinxbase

  2. Download the sphinxbase package from here: https://sourceforge.net/projects/cmusphinx/files/sphinxbase/5prealpha/

  3. Set up sphinxbase by following the tutorial here. - Do not install from the github source files

  4. Install pocketsphinx

  5. Download the pocketsphinx package from here: https://sourceforge.net/projects/cmusphinx/files/pocketsphinx/5prealpha/

  6. Set up pocketsphinx by following the tutorial here.

  7. Set up the speech package

  8. Add the following line to your ~/.bashrc file: - export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib

  9. "Install" the package by adding a link to the speech package from your workspace. - git clone the hlpr_speech repo into ~/Software - e.g. ln -s ~/Software/hlpr_speech/ ~/YOUR_WORKSPACE_NAME/src/

  10. Run catkin_make from your workspace folder.

  11. Creating a dictionary for the speech commands

  12. The speech_listener and speech_gui have a parameter where you specify a .yaml file containing the mapping between keywords and speech commands. The default is kps.yaml file, or if you make your own .yaml file put it in the hlpr_speech_recognition/data/ directory.

  13. The speech_recognizer makes use of pocketsphinx and needs three files in the hlpr_speech_recognition/data directory. First is a list of each speech phrase to be recognized in a text file, this is expected to be kps.txt. If you want to add new commands, add them here and then do step three to create the other files necessary.

  14. Go to the following link and generate the .dic and .lm file. Replace the previous files with the new ones and rename them 6858.dic and 6858.lm respectively.

    • http://www.speech.cs.cmu.edu/tools/lmtool-new.html

Running speech recognition

  1. Running speech with mic input: roslaunch hlpr_speech_recognition speech_rec_w_mic.launch

  2. Running speech with gui input: roslaunch hlpr_speech_recognition speech_rec_w_gui.launch

***** NOTE: Only launch speech_recognizer or speech_gui. Do not launch them at the same time as they have the same functionality

Testing speech recognition

See the main hlpr_documentation repository for information on getting started with the Speech Testing GUI.

Debugging

If either rosrun command doesn't work, then try the following:

  1. Check if your PYTHONPATH is set up right.

  2. Go to hlpr_speech_recognition/bin/ folder and make the nodes executable by typing:

  • chmod u+x speech_recognizer
  • chmod u+x speech_listener
  • chmod u+x speech_gui

Clone this wiki locally