Abstract
The User Experience Challenge: During real-time audio calls, severe background noise or strict social constraints often force the sender to abandon live speech. Currently, users must either endure broken, unintelligible audio or abruptly terminate synchronous calls—destroying conversational flow and stripping away the sender's personal vocal presence from the receiver.
The Technical Solution: This disclosure introduces an on-device AI voice bridge to preserve the sender's uninterrupted vocal presence. The system continuously monitors the call's audio quality at the sender's terminal using multidimensional acoustic measurements. The interface executes a graceful handover to a synthesized stream when environmental noise exceeds critical thresholds or when the sender triggers the transition via alternative inputs, such as whispering.
It instantly activates a safe, multimodal input surface tailored for the sender—accommodating tactile entry, gesture recognition, and silent speech—while seamlessly handing over transmission to an on-device synthesized voice stream. This stream perfectly mimics the sender's unique vocal timbre and conversational pacing. To prevent the receiver from hearing a voice that lacks the natural background "presence" of a live call, the system continuously blends a rolling loop of the sender's environment beneath the synthetic speech. This architecture achieves significant improvements in intelligibility for the receiver while maintaining the sender's vocal identity.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Pujari, Purvam, "Adaptive Communication Interface with Vocal Identity Persistence", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10279