Hearing aid signal-processing system

Claim 2

A signal-processing system for use in hearing aids, comprising: means for producing an electrical signal from human speech and ambient noise; means for classifying said electrical signal into a plurality of non-overlapping frequency bands, which plurality contains a highest frequency band, a next-highest frequency band and a lowest frequency band; means for estimating an absolute quantity of noise in each frequency band; and means for individually varying, as a function of noise estimated by said estimating means, audio gain in each frequency band except the highest frequency band, said varying means comprising means for comparing an absolute quantity of noise in each frequency band except the highest frequency band with an absolute quantity of noise in the next higher frequency band.

Claim 3

The system of claim 1, wherein said estimating means comprises means for establishing an amplitude histogram in the frequency band in which the absolute quantity of noise is to be estimated, and further comprises means for comparing an average signal amplitude with a predetermined quantity and means for setting said absolute quantity of noise at said average signal amplitude when said average signal amplitude exceeds said predetermined quantity.

Claim 4

The system of claim 1, wherein said varying means further comprises means for determining whether the absolute quantity of audible noise in a current frequency band in which gain is to be adjusted exceeds the absolute quantity of audible noise in a next higher frequency band by more than a predetermined threshold value.

Claim 5

The system of claim 3, wherein said varying means operates in a manner that when the absolute quantity of audible noise in a current frequency band in which gain is to be adjusted exceeds the absolute quantity of audible noise in a next higher frequency band by more than said predetermined threshold value, said gain is reduced.

Claim 6

The system of claim 4, wherein said varying means operates in a manner that when the absolute quantity of audible noise in a current frequency band in which gain is to be adjusted exceeds the absolute quantity of audible noise in a next higher frequency band by more than said predetermined threshold value, said gain is reduced to a value at which the absolute quantity of noise in the current frequency band is approximately as loud as the softest speech sounds which can be perceived in said current frequency band.

Claim 7

The system of claim 4, wherein said threshold value is 3 dB.

Claim 8

A signal processing system for use in hearing aids, comprising: means for converting incoming audio information into digital signals; means for classifying said signals by frequency and by amplitude into a plurality of non-overlapping frequency bands and a plurality of non-overlapping amplitude bins within each frequency band, said plurality of said frequency bands including a highest frequency band, a next-highest frequency band and a lowest frequency band; means for establishing an amplitude bin histogram for each of the frequency bands; means for estimating an absolute quantity of audible noise in each frequency band, based upon a corresponding one of the histograms; means for comparing said absolute quantity of audible noise in each band except the highest frequency band with the absolute quantity of audible noise in the next-higher frequency band and determining whether the absolute quantity of audible noise in the lower frequency band exceeds the absolute quantity of audible noise in said next higher frequency band by more than a predetermined threshold value; means for individually varying audio gain of signals in each frequency band; and means for operating said varying means in a manner that when, for a particular frequency band, said absolute quantity of audible noise in the lower frequency band exceeds the absolute quantity of audible noise in said next higher frequency band by more than said predetermined threshold value, said audio gain in said particular frequency band is reduced.

Claim 9

A signal processing system for use in hearing aids, comprising: means for converting incoming audio information into digital signals; means for classifying said signals by frequency and by amplitude into a plurality of non-overlapping frequency bands and a plurality of non-overlapping amplitude bins within each frequency band, said plurality of said frequency bands including a highest frequency band, a next-highest frequency band and a lowest frequency band; means for establishing an amplitude bin histogram for each of the frequency bands; means for estimating an absolute quantity of audible noise in each frequency band, based upon a corresponding one of the histograms; means for comparing said absolute quantity of audible noise in each band except the highest frequency band with the absolute quantity of audible noise in the next-higher frequency band and determining whether the absolute quantity of audible noise in the lower frequency band exceeds the absolute quantity of audible noise in said next higher frequency band by more than a predetermined threshold value; means for individually varying audio gain of signals in each frequency band; and means for operating said varying means in a manner that when, for a particular frequency band, said absolute quantity of audible noise in the lower frequency band exceeds the absolute quantity of audible noise in said next higher frequency band by more than said predetermined threshold value, said audio gain in said particular frequency band is reduced.

Claim 10

The system of claim 7, wherein said estimating means includes means for measuring an average amplitude of said signals over a relatively short period of time and setting said absolute quantity of noise equal to said average amplitude when said average amplitude exceeds a predetermined maximum headroom amplitude.

Claim 11

The system of claim 7, wherein said establishing means further comprises means for decaying said histograms over time.

Claim 12

The system of claim 9, wherein said decaying means operates in a manner that each histogram represents 4 seconds of audio information.