The system output, which uses TTS as presented in Chapter 2, Text-to-speech synthesis could be enhanced by including the use of different voices, languages, and modalities, for example, to present information that is less suitable for spoken output.
The user input techniques for speech recognition that were introduced in Chapter 3, Speech Recognition could be developed in various ways:
By including the similarity measures introduced in Chapter 4, Simple Voice Interactions to compare the recognized results with similar words that might have been what the user actually said
By making use of other techniques for enhancing the recognition results, such as stemming, as described in Chapter 17 of the book Professional Android Sensor Programming by Greg Milette and Adam Stroud
By incorporating confidence measures to support decisions on whether to use confirmations, as described in Chapter 4, Simple Voice Interactions
By making use of the n-best...
