Abstract
The disclosed system and method match verbosity of a machine's responses to the verbosity or brevity of the user's query. The system includes a semantic parser connected to an audio input and output device to process and respond to queries. The method includes building a simple statistical model of mean (M) and standard deviation (SD) for the lengths of the audio of the user’s query utterances and the multiple variants of the generated text-to-speech (TTS) output for each query type or action or intent. The system may extract fluff or slot filling words from verbose queries to use in formulating the response. The system then matches the M and SD between the query and the response to pick an optimum response. The system is more conversational and more dynamically reactive to the user's input. This system retrieves and presents relevant information faster and may be more user-friendly for accessibility users.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Kracun, Aleksandar and Waters, Austin, "Conversational System Responses", Technical Disclosure Commons, (June 09, 2017)
https://www.tdcommons.org/dpubs_series/548