Searcharxiv⌕ Search

arXiv subjects

Juliana Francis

Publications and source records attributed to Juliana Francis.

2 recordsLinked to original sources

Lived Experiences of Power and Agency: Achieving Voice Sovereignty in Assistive Speech Technology for Nonbinary Users

Speech-generating devices (SGDs) are essential tools for many speech-disabled individuals, yet current systems often reinforce normative assumptions about identity and communication. Prior research on gender expansive SGD users has largely focused on gendered voice options, with less attention to broader identity alignment and the technological conditions that shape users' agency. This study examines how gender expansive SGD users experience power over their voices using thematic analysis of qualitative survey data. Three themes emerged: (1) integration of SGDs into daily life, (2) identity expression, and (3) power and agency. Participants expressed the need for more responsive and reliable technology, inclusive and customizable voices, and greater autonomy in defining how their voices sound and are perceived. Ultimately, their perspectives call for a reimagining of SGD systems as tools for reclaiming the power of one's voice and the ability to define and enact one's will through it.

cs.CY↗

No-Shot Text-to-Speech: Limitations of Zero-Shot TTS and its Evaluation Methods in Representing Queer and Transgender Voices

Gender expansive voices remain underrepresented in speech datasets, which could cause challenges with generalizability in zero-shot Text-To-Speech (TTS) systems. We evaluate six zero-shot TTS models on two datasets: one containing non gender expansive voices (N-GE) and one with gender expansive (GE) voices from self-identified speakers. Human ratings found that 2 of the 6 models synthesize less similar GE voices compared to N-GE voices, and 3 have improved performance for GE voices. Automated evaluations yield conflicting results, with speaker similarity and automated mean opinion score predictors rating performance of GE voices inconsistently between models and human ratings. This highlights a potential blind spot in evaluation pipelines, emphasizing a need for more representative data and metrics that better capture human perceptual judgments. Our findings stress the need to include transgender, queer, and minority communities in creation of better datasets and model training.

cs.SD↗