Electrical audio signal processing systems and devices – Hearing aids – electrical
Reexamination Certificate
2000-07-10
2003-06-17
Tran, Sinh (Department: 2643)
Electrical audio signal processing systems and devices
Hearing aids, electrical
C381S316000, C381S320000
Reexamination Certificate
active
06580798
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to a method of suppressing interfering signals in an input audio signal stream, and to a circuit for carrying out the method. The invention is suitable in particular for the improvement of the intelligibility of speech through the suppression of interfering noise in hearing aids or hearing devices.
BACKGROUND
For quite some time, it has been possible with conventional hearing aids to put people with impaired hearing in a position to understand speech, which is spoken in a quiet environment, well again. Difficulties occur, however, when the acoustic environment is full of interfering noise. Over and above the failure to understand speech, the wearers of hearing aids frequently complain that, in such situations, their devices produce an audio output signal which, for them, is unpleasantly loud. For the hearing situation involving interfering noises, various manufacturers have built, for this reason, in addition to using directional microphones, into their latest hearing devices, systems for the suppression of the interfering noise into their latest hearing devices.
Known in this connection are, e.g., the devices SENSO from Widex of Denmark, and PRISMA from Siemens AG of Germany. Both devices are characterized by a processing of the acoustic signal in several separate frequency bands. In the individual partial bands, an examination as to the presence of interfering noise takes place, and depending on the extent of the presence of interfering noise, the affected partial signals are correspondingly attenuated to a greater or lesser degree prior to their renewed re-assembly into a complete signal. The number of the frequency bands in the devices mentioned is limited to three or to four, respectively.
In a joint effort, the companies Resound of the U.S. and Danavox of Denmark have developed digital hearing aids, which are characterized by a processing of the acoustic signal in segments successive in time by means of Fast Fourier Transformation (FFT). The suppression of interfering noise, in the case of these devices, is based on fourteen frequency, bands, which according to the indications of the manufacturers, however, overlap to a great extent. Because of the low level of clarity of modulation when utilizing a maximum of only four frequency bands, resp., because of the great overlap of the fourteen frequency bands calculated from a Fourier transformation, the processes known LIS up to now for the suppression of interfering noise are, in essence, considered merely measures solely for making the output sound of the hearing devices more pleasant. They, however, hardly make any contribution to the objective improvement of the intelligibility of speech. As undesirable side effects, the processing in segments in addition produces a signal delay of more than 10 milliseconds.
It is an object of the present invention to provide a device and a method for the suppression of interfering noise, which objectively improves the intelligibility of speech and, apart from this, manifest an only short time delay (for example, less than 2 milliseconds) between the input signal and the output signal. The object is achieved by the circuit in accordance with the invention and the method disclosed below.
The point of departure for the invention is formed by the American National Standards Institute (ANSI) document S3.5-1997, “Methods for the Calculation of the Speech Intelligibility Index.” In accordance with this standard document, for sufficiently well defined hearing situations, a numerical index value S can be calculated, which assumes real values between zero and one. It provides information about which proportion of the characteristics for speech intelligibility contained overall in spoken speech is accessible to a listener for the comprehension process in the brain in the given situation. For the specific results of a speech test, furthermore the degree of difficulty of the speech material as well as the linguistic competence of the listener are of significance. The decisive point, however, is that the test result in any case proves to be a monotonically increasing function of the index value S.
For the calculation of the index value S, the standard document indicates differing variants, which in the main differ with respect to the number of frequency bands, in which the speech—and noise signals are analysed. The minimum amounts to six bands and the maximum
21
. In every variant, for each frequency band I a value A
i
for the audibility is established, and the index results as weighted sum
S
=
∑
i
I
i
·
A
i
,
Equation
(
1
)
whereby I
i
designate constant, relative significance weightings (importance) for the individual partial bands, i.e., the sum of all these weightings amounts to one.
The values A
i
for the audibility for their part result as products
A
i
=L
i
·K
i
, Equation (2)
whereby L
i
distortion values (distortion levels) and K
i
represent so-called temporary variables, into which the levels of the speech—and of the noise signal enter.
The distortion levels L
i
are calculated in accordance with
L
i
=1−(
E
i
−U
i
−10)/160, Equation (3)
whereby U
i
designate the levels of normal speech defined in the standard document, while E
i
represents the level of the speech signal in the investigated hearing situation.
The temporary variables K
i
finally are calculated in accordance with
K
i
=(
E
i
−D
i
+15)/30, Equation (4)
whereby D
i
signify the levels of an interfering noise and the variables K
i
in all cases are limited to values between zero and one. In a quiet acoustic environment, the values D
i
result as levels of a fictitious interfering noise, which in general are determined by the hearing threshold values of people with normal hearing, resp., in the particular case by those of the individual person with a hearing impairment. In an acoustic environment with a considerable interfering noise, the values D
i
, however, are determined by the external interfering noise plus in addition any masking effects, which are also caused by the interfering noise, by, however, its proportions in bands of lower frequency levels.
From the Equations (1) to (4) it evolves, that two conditions are necessary for achieving the maximum index value S=1. First of all the levels of the speech signal have to be at least 15 dB above those of the interfering noise in all frequency bands. And secondly, in no band must the level of the speech signal be more than 10 dB above that of normal speech U
i
in accordance with the definition of the standard document.
While in a quiet acoustic environment the speech levels in Equation (4) by means of amplification within a hearing device can be raised above the hearing threshold values D
i
of a person with impaired hearing and therefore the temporary values K
i
maximized, the situation under interfering noise is far less favourable. In this case, the amplification raises the levels of the speech signal and of the interfering noise to the same degree, and as soon as the latter exceed the hearing threshold values of the person with impaired hearing they are decisive for the values D
i
, and any further increase of the temporary variables K
i
is therefore impossible.
Simultaneously the level values E
i
under these circumstances as a rule are significantly above those of normal speech U
i
. With this, however, the prerequisites for an increasing of the index value S are also given in the interfering noise, this namely by a reduction of the amplification, as long as the distortion levels L
i
as a result of this once again approach the ideal value 1 and at the same time the temporary variables K
i
remain constant. A further desirable effect in addition results through the diminishing of masking effects, when the interfering noise has significant proportions in low frequency bands, which in practice is often the case.
SUMMARY OF THE INVENTION
The device in accordance with the invention, in parti
Bernafon AG
Oliver Milton
Tran Sinh
Ware Fressola Van Der Sluys & Adolphson
LandOfFree
Hearing aid does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Hearing aid, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Hearing aid will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3159439