We introduce a core for a digital hearing aid that compensates the signal spoken in sensorineural impaired listeners with object of improving their intelligibility. The technique implemented is based on a digital analysis/synthesis of speech: we divided the input signal into short time blocks then we make a multiband analysis, non linear amplification and synthesis based in a sinusoidal model of the voice, according to the subject's dynamic range in each band. The system works in real time and has been implemented with only one ASIC in 1μ ES2 technology including 3 RAM memories with a capacity of 2432 bits and one 16×16 multiplier. The size of the die is 30.59 mm↑（2）.
展开▼
