Use speech_recognition python library

Please use a sd card as extroot to provide enough storage, see https://github.com/respeaker/get_started_with_respeaker/blob/master/QuickStart.md#use-sd-card-to-extend-storage

Speech recognition module for Python is a very useful library to test a variety of Speech To Text (STT) engines:

• CMU Sphinx (works offline)
• Google Speech Recognition
• Wit.ai
• Microsoft Bing Voice Recognition
• api.ai
• Houndify API
• IBM Speech to Text

For ReSpeaker, you can install it by typing

pip install https://github.com/xiongyihui/speech_recognition/archive/respeaker-lite.zip

The package removed some unused files and is fit for ReSpeaker's limited storage.