A new algorithm for frequency transposing in digital hearing aids

Abstract

Frequency transposition is a signal processing strategy of transposing high frequency acoustic content in a speech signal to a low frequency range. This technique has been widely discussed for a long period of time as the conventional amplification strategies are not beneficial to those suffering from the specific condition of high frequency hearing loss. But still, this strategy as a corrective option, never gained a good momentum or interest amongst audiologists as most of the technological solutions developed have got one or the other drawback. The present study was aimed at developing a new algorithm exploring frequency transposition and testing its effectiveness both objectively and subjectively. The developed algorithm involves transposing the significant high frequency acoustic content based on an adaptable threshold, to a specific region in low frequency band, frame by frame. The algorithm also incorporates a compression strategy to adjust the transposed speech content to the hearing loss pattern of the client. The new algorithm eliminates most of the limitations of the existing technologies. Objective evaluation is done with the help of three signal processing software and subjective evaluation with five subjects of normal hearing and ten subjects of moderate to severe sloping hearing loss who are native Kannada speakers within the age range of 20 to 80. 100% detection score is obtained for all the five normal hearing subjects and in high frequency hearing impaired subjects a significant 22% increase is achieved with transposition. The score obtained sufficiently prove that the transposed words have satisfactory sound quality and transposition has increased the speech cues for detection of words. The results hold good for further research in this area and offers a potential rehabilitation technology for persons with high frequency hearing impairment.