In the absence of normal aural feedback, the speech skills of individuals who become deaf in adulthood deteriorate over time, unless they are periodically retrained. Voice Prism» software under development by Language Vision Inc. (LVI) provides an intuitive, real-time, acoustic-based, visible representation of speech sounds for learning these skills. This allows visual access to the nature of speech, and fosters an awareness of the orosensory cues needed for the accurate vowel productions necessary for intelligible speech. We propose to extend this software with a real-time high-resolution spectrogram capability to provide enhanced visualization of consonant information to supplement the existing displays. We hypothesize that our innovative approach to articulation training is far superior to non-instrumental approaches long established and currently used for speech articulation training of the hearing impaired. To test this hypothesis, we also propose to evaluate the efficacy of the resulting software to compare the outcomes of vowel production training with traditional (non-instrumental) vs. those resulting from training with the enhanced software application. These objectives are crucial to LVI's goal of developing self-directed speech training software for individuals with adult onset deafness and will help to determine directions for further development in Phase II work.Summary of anticipated results and implications:The novel and innovative LVI approach represents vowel articulation using intuitive colors and animated graphical tongue and oral cavity representations derived from the speech signals and displayed in real-time. We intend to extend our existing algorithms to produce laboratory prototype software that will run on a standard multimedia PC. The addition of spectrographic capabilities is essential for training in which vowels and consonants are represented in the context of words. phrases, and sentences. Such connected speech training is needed to improve speech intelligibility in real-life situations. Our ultimate goal is to produce products and technologies that provide self-directed, independent learning of speech by individuals with hearing impairments. These products will allow for active learning of speech skills and provide a highly cost-effective means of speech improvement and maintenance. The proposed efficacy study will determine whether visual models and feedback enable hearing-impaired children to learn to accurately produce vowels more rapidly than with traditional methods. It is predicted that children who receive training with the enhanced LVI software will make more rapid gains in vowel production and generalize their learning to unpracticed words to a greater extent than peers who receive traditional intervention. This study will also afford opportunities for participating speech-language pathologists to offer their suggestions for software modification based on clinical observation and experience.
