Patients with severe hearing loss have the option to get a cochlear implant device to regain their hearing. Yet, the implantation process is not always optimal, which in some cases results in a shallow insertion depth...Patients with severe hearing loss have the option to get a cochlear implant device to regain their hearing. Yet, the implantation process is not always optimal, which in some cases results in a shallow insertion depth or an accidental insertion into the wrong cochlear duct. As a consequence, the patients' pitch discrimination ability is suboptimal, leading to an even more decreased vowel identification, which is vital for speech recognition. This paper presents a technical approach to solve this problem: the adaptive pitch transposition module modifies the frequency content in a fashion so that the pitch is fixed to an optimal value. To determine this value, a patient-individual best pitch is determined experimentally by evaluating speech recognition at different pitches. This best pitch is subsequently called the comfort pitch. As a result of the considerations a technical implementation is presented in principle. A system comprised of pitch detection, pitch transposition and an arbitrary chosen comfort pitch is described in depth. It has been implemented prototypically in Matlab/Octave and tested with an example audio file. The system?itself is designed as a preprocessing stage preceding cochlear implant processing.展开更多
文摘Patients with severe hearing loss have the option to get a cochlear implant device to regain their hearing. Yet, the implantation process is not always optimal, which in some cases results in a shallow insertion depth or an accidental insertion into the wrong cochlear duct. As a consequence, the patients' pitch discrimination ability is suboptimal, leading to an even more decreased vowel identification, which is vital for speech recognition. This paper presents a technical approach to solve this problem: the adaptive pitch transposition module modifies the frequency content in a fashion so that the pitch is fixed to an optimal value. To determine this value, a patient-individual best pitch is determined experimentally by evaluating speech recognition at different pitches. This best pitch is subsequently called the comfort pitch. As a result of the considerations a technical implementation is presented in principle. A system comprised of pitch detection, pitch transposition and an arbitrary chosen comfort pitch is described in depth. It has been implemented prototypically in Matlab/Octave and tested with an example audio file. The system?itself is designed as a preprocessing stage preceding cochlear implant processing.
