Package for speech recognition using Google's Speech API



  • This package allows the recognition of speech using Google's speech engine. The audio is sampled and send to the Google's servers, which in turn will process it and return the speech content and the confidence value of the recognition. It has an advantage of speaker indepedent recognition and no training required. The disadvantage is that it a bit slower unless you have a good internet connection.

    The node publishes two topics: /speech and /confidence. /speech is the recognized speech which is of the standard String message type./confidence is the confidence level of recognition which is of the standard Int8 message type.

Quick Tutorial

  • sudo apt-get install sox
  • clone the git
  • rosrun gspeech

Wiki: gspeech (last edited 2012-06-30 17:58:34 by achuwilson)