Beyond Conventional Speech-Rate Control: A Unified Framework for Bidirectional Adaptive Speaking Rate Regulation Between Generative AI and Human Users
Speaking rate is a fundamental dimension of verbal communication that influences intelligibility, information processing, conversational rhythm, perceived naturalness, and interaction efficiency. Although speech-rate control has been extensively investigated in speech synthesis, spoken dialogue systems, and human–computer interaction, existing approaches have predominantly addressed speech generation parameters, system-side adaptation, or users' preferences for machine-generated speech. These research directions do not necessarily provide a unified account of speaking-rate regulation as an interactional capability in which both a human user and a generative artificial intelligence system can initiate, modify, negotiate, and continuously regulate speaking speed. This paper introduces the Bidirectional Adaptive Speaking Rate Control (BASRC) Framework, a proposed conceptual framework that treats speaking rate as a dynamically configurable and mutually regulatable dimension of generative AI communication. The framework distinguishes user-initiated AI rate control, AI-initiated self-regulation, user-side rate control, AI-guided user rate adjustment, bidirectional rate regulation, adaptive synchronization, contextual rate adaptation, and negotiated speaking-rate selection. It further introduces a layered architecture encompassing rate preference representation, interactional state assessment, control authorization, rate decision-making, temporal speech realization, feedback monitoring, and preference persistence. Unlike conventional approaches that primarily treat speaking rate as a fixed synthesis parameter or a one-directional adaptation variable, BASRC conceptualizes it as an interaction-level control capability involving multiple agents, control directions, temporal scopes, and decision authorities. A comparative analysis of existing literature identifies the theoretical foundations and technical precedents of speech-rate control while establishing the proposed framework's intended contribution at the level of unified interaction architecture. The paper does not claim that individual speaking-rate adjustment mechanisms have never existed. Rather, it advances a new framework-level research proposition: a comprehensive model for representing and coordinating human–AI speaking-rate regulation across multiple control configurations. This work establishes a foundation for future implementations and empirical investigations of more flexible, personalized, and mutually adaptive spoken communication.
Authors
- Jincheng Zhang (ORCID: https://orcid.org/0009-0005-1860-0009)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-17
- DOI
- https://doi.org/10.5281/zenodo.24298176
- Primary Topic
- Speech and dialogue systems
- Type
- preprint