Speech Acronym Identifier

Android application which can pick up Acronyms from general speech and provide their definitions.

My contribution:

  • Developed intent recognition and surfacing of acronym translation

 

There are many technical acronyms used by the speakers that clients might not understand. The main objective of this feature is to be used during the speaker sessions. The microphone on the clients’ phone will recognize what the speakers are talking about and if there is an acronym mentioned, the system will display its definition on the screen.

The acronym identifier is composed of two parts: a python script and an Android application.

The python script is used to train the SpeechToText (STT) service to recognize custom words, in this case acronyms which are not a part of the standard English dictionary. So this script simply sends a list of words and their pronunciations to STT to train it and obtain a workspace ID. At the time of the writing the STT custom grammar feature was still in beta and not published.

The second part is the standalone Android application itself. The platform was chosen as there were existing libraries for STT and the ease of development. Once opened the user can tap on the screen to start the acronym recognition and long press to stop. Once started the microphone is turned on and the stream of the data is sent to the STT service. The definition of the acronym is displayed on the screen for a brief period of time.

Leave a comment