ISCA Archive Interspeech 2024
ISCA Archive Interspeech 2024

VoxFlow AI: wearable voice converter for atypical speech

Grzegorz P. Mika, Konrad Zieli´nski, Paweł Cyrta, Marek Grzelec

The central challenges of the embedded voice conversion system lie in the transformation of atypical speech into one resembling the original voice without changing the linguistic content. While various atypical speech input techniques have been developed to aid voice inter- actions, they are typically challenging to use daily, and the research on their usability could be more robust. In this Show&Tell paper, we present VoxFlow AI, a wearable loudspeaker integrated with a voice converter designed to improve real-life interactions. During the interactive demonstration, we will show how this device changes interactions between people with oncological and neurological speech disorders, who are the direct beneficiaries of the solution.