Searcharxiv⌕ Search

arXiv subjects

Paulo A. Condado

Publications and source records attributed to Paulo A. Condado.

4 recordsLinked to original sources

DIY Customized Speech Recognition for a Person with Speech Disabilities

General-purpose state-of-the-art automatic speech recognition (ASR) systems fail systematically for speakers with moderate to severe speech disabilities. This paper describes the design, implementation, and evaluation of a customized ASR system built for the first author, a researcher with cerebral palsy and dysarthria. We fine-tuned OpenAI's Whisper medium and large-v3 models using Low-Rank Adaptation (LoRA) on audio samples recorded from the first author's own speech on consumer-grade hardware. On a held-out test set of 100 previously unseen utterances, the corresponding transcriptions resulted in a Word Error Rate drop from 49% to 17% for medium and from 30% to 12% for large-v3, while the Character Error Rate fell from 26% to 7% for medium and from 16% to 5% for large-v3, indicating that the residual errors are minor phonemic substitutions. Based on these models, the first author developed AVoz, a mobile application that acts as a real-time interpreter, converting his speech into synthesized voice for the listener. All processing is performed locally, ensuring that no audio data leaves the user's machine. We argue that the methodology is replicable. Moreover, we publish the full LoRA configuration, the implementation, and a step-by-step recording and training protocol, so that other people with speech disabilities, or their family members and caregivers, can build their own systems with a few hours of labeled audio data and a couple of days of GPU time.

cs.SD↗

Evaluating Accessible Synchronous CMC Applications

This paper proposes a more comprehensive evaluation methodology to measure the usability and user experience qualities of accessible synchronous computer-mediated communication applications. The methodology goes beyond current practices by evaluating how the interaction between a user and a product influences the user experience of those at the other endpoint of the communication channel. A major contribution is given with the proposal of a user test where one of the participants tries to guess whether the other participant has a disability or not. The proposed test is inspired in the Turing Test, and is a consequence of user requirements elicited from a group of individuals with motor and speech disabilities. These ideas are tested and validated with two examples of synchronous communication applications.

cs.HC↗

EasyVoice: Integrating voice synthesis with Skype

This paper presents EasyVoice, a system that integrates voice synthesis with Skype. EasyVoice allows a person with voice disabilities to talk with another person located anywhere in the world, removing an important obstacle that affect these people during a phone or VoIP-based conversation.

cs.CY↗

Breaking barriers for people with voice disabilities: Combining virtual keyboards with speech synthesizers, and VoIP applications

Text-to-speech technology has been broadly used to help people with voice disabilities to overcome their difficulties. With text-to-speech, a person types at a keyboard, the text is synthesized, and the sound comes out through the computer speakers. In recent years, Voice over IP (VoIP) applications have become very popular and have been used by people worldwide. These applications allow people to talk for free over the Internet and also to make traditional calls through the Public-Switched Telephone Network (PSTN) at a small fraction of the cost offered by traditional phone companies. We have created a system, called EasyVoice, which integrates speech synthesizers with VoIP applications. The result allows a person with motor impairments and voice disabilities to talk with another person located anywhere in the world. The benefits in this case are much stronger than the ones obtained by non-disabled people using VoIP applications. People with motor impairments sometimes can hardly use a regular or mobile phone. Thus, the advantage is not only the reduction in cost, but more important, the ability to talk at all.

cs.CY↗