Abstract
Many conversations between individuals include recommendations of various items (books, movies/shows, concerts, etc.) as well as services (e.g., restaurants). Remembering such recommendations requires users to jot them down during or after the conversation. In contrast, visual items such as posters, or items displayed on a screen can be captured easily via a device camera or as screenshots. This disclosure describes techniques that enable, with the permission of participants in a conversation, the use of device microphones to automatically obtain conversational audio that is analyzed with an on-device artificial intelligence (AI) model to extract recommendations. Analogous to long-pressing for screenshots, users can long-press a combination of buttons to capture an audio recommendation. The AI model can output the identified recommendations in easily digestible formats, such as maps (with points of interest or directions), playlists (for media content), steps in a recipe, etc.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Marques, Paula, "Audioshot: Seamless Capture of Word-of-Mouth Recommendations", Technical Disclosure Commons, (October 10, 2025)
https://www.tdcommons.org/dpubs_series/8703