Native voice, self-concept and the moral case for personalized voice technology

Abstract

Purpose (1) To explore the role of native voice and effects of voice loss on self-concept and identity, and survey the state of assistive voice technology; (2) to establish the moral case for developing personalized voice technology. Methods This narrative review examines published literature on the human significance of voice, the impact of voice loss on self-concept and identity, and the strengths and limitations of current voice technology. Based on the impact of voice loss on self and identity, and voice technology limitations, the moral case for personalized voice technology is developed. Results Given the richness of information conveyed by voice, loss of voice constrains expression of the self, but the full impact is poorly understood. Augmentative and alternative communication (AAC) devices facilitate communication but, despite advances in this field, voice output cannot yet express the unique nuances of individual voice. The ethical principles of autonomy, beneficence and equality of opportunity establish the moral responsibility to invest in accessible, cost-effective, personalized voice technology. Conclusions Although further research is needed to elucidate the full effects of voice loss on self-concept, identity and social functioning, current understanding of the profoundly negative impact of voice loss establishes the moral case for developing personalized voice technology. Implications for Rehabilitation Rehabilitation of voice-disordered patients should facilitate self-expression, interpersonal connectedness and social/occupational participation. Proactive questioning about the psychological and social experiences of patients with voice loss is a valuable entry point for rehabilitation planning. Personalized voice technology would enhance sense of self, communicative participation and autonomy and promote shared healthcare decision-making. Further research is needed to identify the best strategies to preserve and strengthen identity and sense of self.

Perceptual prioritisation of self-associated voices

“…Synthesized voices provided by augmentative and alternative communication (AAC) devices are often unnatural-sounding, lacking in emotional prosody and slower than self-produced speech. AAC users report feeling like they present a distorted representation of themselves, which can decrease the desire to engage in social interaction and contribute to the loss of a social identity (Nathanson, 2017). Only recently has the moral case for more personalised voices been considered, with companies such as VocaliD (www.vocalid.ai) now providing a selection of voices from which AAC users might choose a voice that better conveys their self-identity.…”

Innovations for Supporting Communication: Opportunities and Challenges for People with Complex Communication Needs

“…As Alan Martin [55], a long-time user of an SGD, commented, “Although I now have an English male voice, I sound more like a BBC news reader than the ‘Scouser’ that is the real ME” (p. 99). The aspiration of capturing “the unique nuances of individual voice” [47, p. 73] so that individuals can express their identity and use voice as “an embodiment of self in a social context” (p. 74) is not yet within reach.…”

“…This database has been generated by volunteer donations of over 20,000 speakers representing a diversity of age, gender, cultures, and geography [48, 50]. As this innovation becomes more widely available, the future could (and should [47]), see unique and personalized voices as the norm.…”

“…In parallel to the innovations in hardware for SGDs and in access options, innovations in voice technology have transformed the range of synthetic voices now available, leaving behind the robotic, monotone voice output of early devices, through advances in processes for capturing and reconfiguring voice data [47]. Synthetic speech has found its way into routine technologies, including navigation systems, gaming, and telephone-based support services for a wide range of industries, (e.g., utility companies, banking).…”

“…Currently, corpus-based synthetic voices (even those generated from very large corpora such as the Human Voicebank TM ) still lack the potential to generate the intonation and prosody that characterizes natural speech [47]. There is growing impetus, both commercially and in relation to SGDs, to re-create variation in voice qualities such as tone of voice, to capture not only the linguistic content of utterances, but also a speaker’s intended meaning and vocal identity [48, 52-54].…”