JP3151623B2 - Noise suppression circuit - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to suppression of noise during signal transmission, and more particularly to background noise suppression for separating voice and voice-like signals in an audio signal from background noise.

BACKGROUND OF THE INVENTION Various types of background noise accompany speech in audio communication systems designed primarily to transmit speech. Such types of background noise include hissing of air conditioning systems, cars, other mechanical devices, crowd background noise, and white noise, among others. All of these different types of noise exhibit some temporary property that distinguishes them significantly from sounds produced by a single speaker.

Many conventional noise suppression systems have been developed so far. For example, squelch systems whose primary purpose is to remove channel noise from weak or non-realistic radio (RF) carriers, expansion / compression systems such as DBX and Dolby expanders / compressors, and noise pulse eliminators of various designs It is.

In conventional RF squelch systems, examples thereof are U.S. Pat. Nos. 3,660,765 (Glasser et al.) And 4,638,604 (Koebe).
r) and 4,158,174 (Gruenberger et al.), when useful information is present in the derived audio channel, is closed to transfer that information to the output line;
On the other hand, in weak signal-to-noise conditions, it is usual to set the threshold manually or automatically so that the gate opens and squelches the output signal. Such ordinary squelch systems operate to determine that the received RF signal has a high signal-to-noise ratio,
However, it does not examine the contents of the RF signal to determine whether it is speech or background noise. In contrast, the present invention determines what the content of the obtained audio signal is, and suppresses background noise while leaving the audio signal.

In the case of expander / compressor systems such as DBX and Dolby systems, the emphasis is on examining and adjusting the relative amplitudes of the different components of the audio signal, with the ultimate goal being to reduce background noise. In such systems, no distinction is made between sustained sounds, such as long drum hits, and changing sounds, such as spoken words. The present invention, on the other hand, is directed to distinguishing between sound persistence and non-persistence, such as drum hits being noise, but speech being separated as a desired signal.

In prior art noise pulse eliminator structures, attempts have been made to examine the statistics of a given waveform to see if they are relevant to an RF transmission system or a recording system. However, the nature of the analysis performed in the noise pulse remover is very different from the nature of the analysis performed in the present invention. While the noise burst remover is designed to eliminate fast burst type signals, the present invention is designed to eliminate long-lasting sounds. Inevitably, differences in emphasis will result in differences in the methods and equipment employed to achieve the desired results.

In still other types of prior art Neuss suppression systems, a variation of the conventional RF squelch system is used, in which the audio signal is analyzed and the sound persistence and non-persistence are distinguished in a voice communication system. . In this type of noise suppression system, exemplified by U.S. Pat. No. 4,461,025 (Franklin), an audio signal containing both background noise and audio information is a short-time power having separate energy time lengths therein. Integrate for a period of 20-100 ms to obtain a spectrum. The power spectrum obtained when audio information appears in the audio signal includes an energy period of zero or near zero, but the power spectrum obtained when only background noise appears in the audio signal, if any. It contains few such zero energy levels. This distinction is used to generate a control signal that is only present when only audio components are present in the audio signal. This control signal is applied to a signal amplifier that receives the audio signal as a second input signal.

The multiplier output signal is zero if either of the two input signals is not present. Therefore, there is no multiplier output signal during the time when only the background signal appears in the audio signal. On the other hand, the audio signal containing the background noise is amplified during the time the audio signal appears therein. The present invention is distinguished above in that the background noise portion of the audio signal is removed rather than amplified with the audio portion of the signal.

In view of the foregoing, the present invention provides for the extraction of audio parameters, particularly audio parameters, such as the amplitudes of various portions of the audio signal, and for later re-establishment in, for example, cochlear implants, vocoders, speech recognizers and audio transmission data rate reduction systems. It will be appreciated that in the configured system it is intended to suppress background noise.
For convenience, the present invention will be described with respect to suppressing noise in a voice transmission system used in cochlear implants for severe hearing loss, but the present invention is not so limited and other signal transmission systems and It can be applied to other devices.

For many people who have severe hearing loss, the reason for hearing loss is the absence or damage of hair cells in the cochlea that convert acoustic signals into nerve impulses. These people therefore cannot benefit from ordinary hearing aids for their inability to generate neural stimulation from sound in the normal manner, no matter how large the acoustic signal.

The cochlear implant system seeks to bypass these hair cells by directing electrical stimulation to acoustic nerve fibers and directing sound perception in the brain. Numerous methods have been described in the past for achieving this goal. One of the most effective of these is Crosby et al., US Patent No. 4, entitled "Cochlear Implantation System for Hearing Prostheses."
It is shown and described in 532,930. The subject matter of the aforementioned Crosby et al. Patent is hereby incorporated by reference.

The Crosby et al. Patent describes a cochlear implant system in which an electrode array comprising multiple platinum ring electrodes in a silicone carrier is implanted in the cochlea of the ear. The electrode array is connected to a multi-channel receiver / stimulation unit containing semiconductor integrated circuits and other components, which unit is implanted in the patient adjacent to the ear. The receiver / stimulation unit receives data information and power via the tuning coil via an inductive link with an external audio processor that can be worn by the patient. The audio processor includes an integrated circuit and various components arranged or mapped to emit data from an erasable programmable read only memory (EPROM). EP
The ROM is programmed to suit each patient's perception of electrical stimulation as determined by testing the patient and his implanted stimulator / electrodes. This test is performed using a diagnostic and programming unit (DPU) connected to the audio processor by an interface unit.

The system of Crosby et al. Allows the use of a variety of speech processing strategies, including dominant spectral peaks and amplitude compression of speech pitch to include voiced, unvoiced glottal, and rhythmological information. The voice processing strategy employed is based on known psychophysical phenomena and is tailored to individual patients through the use of diagnostic and programming units. Biphasic pulses are supplied to various combinations of electrodes by switch control current sinks in various modes of operation. The transmission of data is by a series of separate data berths representing the amplitude determined by the selected electrode, electrode mode shape, stimulation current, and the duration of the amplitude burst.

A problem often encountered by cochlear implant recipients is that background noise is irritating and causes unnecessary stimulation. Implants with impaired sound processing ability cannot easily filter or ignore such stimuli from background noise, as in a normal listener. In the past, attempts have been made to solve this problem by using an activation switch that turns the audio processor on when the signal level increases and then turns off the processor when the signal level drops. However, this has the effect that some initial sounds are lost. In addition, sudden on / off switching of the signal is distracting and irritating to the implant recipient. Also, the residual noise in the composite audio signal is provided to the implant recipient along with the audio information therein.

DISCLOSURE OF THE INVENTION According to one aspect of the present invention, there is provided a method for suppressing residual noise in processing a composite input signal having a slowly varying residual noise level component and a rapidly varying data component. The method comprises the steps of: canceling a residual noise level component from the composite signal to the composite input signal; and outputting an output signal having a predetermined polarity and substantially corresponding to a rapidly changing data component of the input signal. Adding a slowly varying signal to provide: sensing the output signal; and providing a means for ensuring that the output signal maintains the predetermined polarity.

According to another aspect of the present invention, a composite input signal having a residual noise level component and a data component is received, the noise level component is eliminated, and a single polarity substantially corresponding to a predetermined polarity and the data component. A noise suppression system for providing a polarity output signal is provided. The system comprises: first means for providing a signal corresponding to a value of the residual noise level component; and providing, from the signal, the unipolar output signal corresponding to the data component of the input signal. Second means for subtracting the composite input signal; and third means responsive to the unipolar signal responsive to ensure that the unipolar output signal maintains the predetermined polarity. .

In accordance with yet another aspect of the present invention, there is provided a noise suppression system for receiving an input signal having a low frequency residual noise level component and a high frequency data component therein. The system circuit includes a first means for providing a voltage corresponding to a voltage of a residual noise level of the input signal, and a circuit for converting the input signal to provide an output signal corresponding to a data component of the input signal. Or a second means for subtracting from the corresponding voltage and limiting the value of the applied voltage, thereby preventing this voltage from exceeding the value of the residual noise level portion of the input signal or the corresponding voltage. A third means responsive to the first means and coupled thereto.
Means.

According to yet another aspect of the present invention, a plurality of the above-described noise suppression circuits are provided. In this case, the input signal is separated into a plurality of different frequency band channels, each having an independent noise reduction circuit. In addition, in various noise reduction circuits, the first means for providing a voltage corresponding to the voltage of each noise level component is to facilitate selective reduction of noise in different frequency band channels. Give different response times.

BRIEF DESCRIPTION OF THE DRAWINGS Other objects, features and advantages of the present invention will become apparent when considering the following detailed description in conjunction with the drawings. In the drawings, FIG. 1 is a circuit diagram of a noise suppression system according to one embodiment of the present invention.

FIG. 2 is a graph showing input signal levels plotted against time for the circuit of FIG.

FIG. 3 is a graph of capacitor voltage plotted against time for the circuit of FIG.

FIG. 4 is a graph of output signal level plotted against time for the circuit of FIG.

FIG. 5 is a circuit diagram of a noise suppression system according to a second embodiment of the present invention.

FIG. 6 is a digital execution flowchart of a noise suppression system according to another embodiment of the present invention.

FIG. 7 is an embodiment of the present invention in which a plurality of noise suppression circuits having different response times are employed to selectively reduce the noise level in an input signal separated into a corresponding number of frequency band channels. It is a block diagram of.

BEST MODE FOR CARRYING OUT THE INVENTION Referring to FIG. 1, a noise suppression system according to one embodiment of the present invention is indicated generally by the numeral 1. An input signal, generally indicated at 10, is applied to system 1 via conductor 11. The input signal 10 comprises a signal having a slowly varying residual noise level component therein and a rapidly varying data component therein. Input signal 10 may comprise, for example, a rectified audio signal generated in the aforementioned Crosby et al. U.S. Pat. No. 4,532,930.

The input signal 10 is connected to a subtraction device 12, for example one of the inputs of a summing amplifier. Capacitor 14 is slowly charged by current source 13 connected in series thereto, and provides a second voltage to subtraction device 12, the difference between the two inputs is provided on line 16 as output signal level parameter 15. Is done.

Line 16 is connected through line 17 to one input of a comparator 18, for example an operational amplifier, the other input being grounded. Comparator 18 serves to sense that output signal 15 is about to go negative. The output of comparator 18 is applied through line 19 to a field effect transistor 20 whose source and drain terminals are connected across capacitor 14. Thus, if the output signal level parameter 15 is going to go negative, the field effect transistor 20 will operate to discharge the capacitor 14 until the output signal parameter 15 becomes positive again.

In operation, if there is constant background noise in the input signal 10, the capacitor 14 will discharge to the background noise level and its voltage will be subtracted from the input signal 10 to provide an output signal level parameter 15. .
In fact, the current source 13, the capacitor 14 and the subtractor 12 are
To virtually remove the residual noise level component from the input,
It comprises means coupled to the input of the noise suppression system 1 for adding a slowly varying negative signal to the input signal 10. Similarly, the comparator 18 serves as a means of sensing the output signal level parameter, and the field effect transistor 20 quickly applies a positive signal to the signal being processed to prevent the output signal from becoming more negative. It serves as a means for responding to the sensing of a negative value of the output signal.

Referring to FIG. 2, which is a graph showing the input signal level plotted against time for the circuit shown in FIG. 1, not the signal itself, but the signal energy, the envelope, the average, or other measurements of the signal level. The parameter representing the input signal expressed in the form of rises and falls when receiving the voice. Between words 40 and 41, the parameter does not fall to zero and the plateau representing the residual noise level or noise floor
Descent to 42. Similarly, between words 41 and 42, the parameter descends to noise floor 44. Audio signal energy fluctuates rapidly over tens or hundreds of milliseconds,
Background noise from, for example, fans, air conditioners, or automobiles fluctuates more slowly over tens of seconds. The noise suppression system of FIG. 1 removes slowly varying noise energy from the signal level parameters and leaves more rapidly varying components.

Referring to FIG. 3 in conjunction with FIGS. 1, 2 and 4, after start-up, it will increase slowly over time as shown as voltage charge 51 on capacitor 14. This charge voltage is subtracted from the input signal 10 until the noise floor represented by plateau 42 (FIG. 2) is effectively removed, as shown at 42a in FIG. Thereafter, the voltage charge on capacitor 14 remains substantially constant, as shown at 52 in FIG.
4 rises slightly between words, as shown at a, falls again at the end of the word, as shown at 52b, and the noise floor is reduced to 44a in FIG. 4 until a change in the residual noise level of the input signal 10 occurs. Continue at substantially zero level as indicated by. By eliminating the noise level, the patient receives little or no stimulation when the audio signal of the signal stops.

If the residual noise level of the input signal drops as shown at 46 in FIG. 2 due to, for example, a stoppage of a fan or the like, such a drop in the noise level floor has the effect of forcing the output signal level 15 to be negative. Would. This causes the field effect transistor 20 to discharge a portion of the voltage on the capacitor 14 as shown at 53 in FIG. 3, so that a new substantially constant low voltage charge is obtained as shown at 54 in FIG.
Appears at 14. This new constant voltage corresponds to a new noise floor plateau 46 (FIG. 2) caused by the change in residual noise level, and the corresponding output signal level floor goes to substantially zero level as shown at 46a in FIG. Return. To this end, the noise suppression system 1 adapts to the reduction of the residual noise level and adjusts to the patient's threshold to provide the lowest signal level output (noise floor plateau 46) above the threshold stimulus.

The time constant for charging the capacitor 14 is
Preferably between about 2 seconds and about 10 seconds. The effect of using a short time constant, eg, about 2 seconds, is that background noise is quickly removed. However, persistent signals will also be removed. Long time constants, for example about 10 seconds, will take longer to remove background noise. However, such a long time constant will not significantly affect the lasting signal. As will be explained in greater detail below, when the system 1 of FIG. 1 is used in a cochlear implant system using multiple electrodes provided with different frequency band portions of each of the input signals, short-term constants are characteristic in speech. It is preferably used for the high frequency band portions that are temporary and the long time constant is used for the low frequency portions where the sound is more sustained. The aforementioned time constant range of 2 to 10 seconds should not be taken as absolute. This is because time constants outside this range also operate to the extent that they change.

Referring now to FIG. 5, an alternative embodiment 2 of the noise suppression system according to the present invention employing a precision diode network is shown generally at 2. In this embodiment, input signal 30 is AC connected through capacitor 31 to operational amplifier 32, and the other input of amplifier 32 is grounded. The output of capacitor 31 is also connected to one side of resistor 33, and the other end is connected to ground. The output of the operational amplifier 32 is connected through a diode 34 to the output line 35 of the noise suppression system 2, and the output signal level parameter 36 appears at the downstream end of the output line 35 during operation of the system.

Resistor 33 serves to slowly and continuously lower the positive output level of the system to bring the residual noise floor to substantially zero. Therefore, this corresponds to the function of the current source 13 and the capacitor 14 in the embodiment of FIG.
On the other hand, the operational amplifier 32 and the diode 34 act to clamp the output if the output signal level parameter 36 is going to drop below zero. If the output signal level 36 is going to drop below zero, the capacitor 31 is quickly charged by the amplifier 32 and the diode 34, restoring the output signal level parameter to a positive value. Amplifier 32
And the diode 18 thus corresponds in function to the comparator 18 and the field effect transistor 20 of FIG.

The circuits of the noise suppression system of FIG. 1 and FIG.
It should be noted that it operates on a slowly changing parameter that represents the signal level, not the signal itself, but its energy, envelope or average, or other measure of the signal level. These systems work equally well regardless of the signal level measurements employed. This system is useful in cases where the signal is analyzed into its level and frequency (or frequency band) components and reconstructed, such as vocoders, certain audio processing hearing aids and cochlear implants. In the case of a cochlear implant, as previously indicated, the signal is analyzed into its parameters and provided to the patient in the form of electrical stimulation. As also indicated earlier, the noise suppression system is applied independently to multiple frequency bands in the case of cochlear implants, so that different noise plateaus at different frequencies are still accurately subtracted even if the noise spectrum is not uniform. be able to.

Referring to FIG. 6, a flow chart representing a digital embodiment of the noise suppression system according to the present invention is generally 3
It is shown as The starting point for this flowchart is 60
It is shown in This represents a conventional initialization step, during which the associated microprocessor (not shown) is initialized when the system starts. After initialization, the noise suppression algorithm directs the system to sample the composite input signal level, as shown at 61. It then increments the noise floor level as shown at 62 and determines if the noise floor level is greater than the composite signal input level as shown at 63. If the noise floor level is greater than the signal level as shown at 64, the noise floor level will be reduced to be equal to the signal level as shown at 65. If the noise floor level was reduced equal to the signal level as shown at 65, or if the decision made at 63 as shown at 66 is not greater than the signal level, the composite signal level minus the noise floor level signal. An equal output signal is generated as shown at 67.
The algorithm then backs out to sample the composite input signal level again at 61, as shown at 68, and repeats itself at a higher rate while the system is operating.

Referring now to FIG. 7, a noise suppression system for providing a plurality of output signals in different frequency bands is shown generally at 4. As previously indicated, multi-frequency systems for use in cochlear implants for severe hearing loss are known. One example of such a system is US Patent No. 4,184,856 to Hochmair et al. Entitled "Multi-Frequency Systems and Methods for Enhancing Auditory Stimulation and the Like". Hochm
The air et al patent discloses a system in which a plurality of carrier signals are modulated by pulses corresponding to signals in an audio frequency band. The carrier signal is transmitted to a receiver having a separate channel for receiving and demodulating the transmitted signal. The detected pulses are applied to electrodes on a prosthetic device implanted in the cochlea. The electrodes are positioned in the cochlea to stimulate an area having a desired frequency response. The pulse has a frequency corresponding to the frequency of the signal in the audible band and a pulse width corresponding to the amplitude of the signal in the audible band. The disclosure of the Hochmair et al. Patent is incorporated herein by reference.

As shown in FIG. 6, the input signal 70 has a total of 71, 72, 73, 7
Applied to each of the plurality of channels indicated by 4 and 75. Each of the channels 71-75 should allow only a limited portion of the frequency band of the input signal to pass through,
Corresponding means for filtering the input signal are included therein. To this end, channel 71 includes a filter means 71a designed to pass a portion of the input signal having a frequency between 300 and 1000 Hz. 1000H for channel 72
It includes a filter means 72a designed to pass a portion of the input signal having a frequency above z. Channel 7
3 includes filter means 73a designed to pass a portion of the input signal having a frequency between 2000 and 2800 Hz. Channel 74 comprises a filter means 74a designed to pass an input signal having a frequency between 2800 and 4000 Hz. Channel 75 comprises filter means 75a designed to pass a portion of the input signal having a frequency between 4000 and 8000 Hz.

Similarly, each of channels 71-75 is
Means are provided for rectifying the frequency band portion of the input signal 70 that has passed through the corresponding filter means 71a-75a. The various rectifying means include 71b, 72b, 73b, 74b and 75b.
Has been identified as The rectifying means 71b to 75b are connected in series with the respective filter means 71a to 75a.

Each of the channels 71-75 also has corresponding means for eliminating residual noise level components from the various rectified input signals in the channels. As described above, the channel 71 is a means for eliminating the residual noise level or the noise suppressor.
The channel 72 includes a residual noise level canceling means or noise suppressor 72c, the channel 73 includes a residual noise level canceling means or noise suppressor 73c, and the channel 74 includes a residual noise level canceling means or noise suppressor 74c. The channel 75 includes a residual noise level canceling means or a noise suppressor 75c. The various noise suppressors 71c-75c have their corresponding filter means 71a-75c and their corresponding rectifier means 71b-75b.
Connected in series. The noise suppressors 71c-75c preferably include any of the noise suppression systems 1 and 2 shown in FIGS. 1 and 5, respectively. In addition, the time constant or response time of the various noise suppression circuits 71c-75c, as indicated above, ranges between 2 and 10 seconds, and for the reasons previously indicated, the higher frequency band channel is It is desirable to have a short time constant and the low frequency band channels have a long time.

From the foregoing, it will be appreciated that the improved noise suppression system described herein removes residual noise in a signal without affecting the provision of data in the signal. Further, such noise suppression systems rapidly compensate for changes in residual noise level or noise floor, and are therefore particularly useful for use in devices used to assist hearing impaired or hearing impaired persons. Furthermore, the present invention provides an improved noise suppression circuit in which the noise component of the input signal can be reduced simultaneously and independently in different frequency bands of a number of input signals.

While particular embodiments of the present invention have been illustrated and described, it will be obvious to those skilled in the art that various other changes and modifications can be made in broad aspects without departing from the invention.
It is therefore intended that the following claims cover all such changes and modifications as fall within the true spirit and scope of the invention.

For example, the output from the system for suppression of residual noise is generally referred to herein as a positive signal, and monitoring and correction means are described as thereby maintaining the output signal positive.

It is anticipated that embodiments of the present invention are also possible in which the window of reference is inverted, whereby the output signal is negative with respect to the reference, and corresponding correction means are used to keep the output progression negative with respect to the reference.

Continuation of the front page (56) References JP-A-53-906 (JP, A) JP-A-60-140399 (JP, A) Yanagisawa, Kanemitsu; Electronic Science Series 52 Active Filter Design, Third Edition (1976) 2.2.20), published by Sanpo Co., Ltd., p. 51-58 (58) Field surveyed (Int.Cl. ⁷ , DB name) H04B 1/10 H04B 15/00 G10L 3/00 H04R 25/00

Claims (28)

(57) [Claims] A system for suppressing residual noise in the processing of an input signal having a slowly varying residual noise component and a rapidly varying data component therein, said input portion receiving said input signal. And the output section for outputting the output signal from the system (16; 3
5) a first means comprising a circuit having: 5) connected to an input portion (11) of the first means for substantially eliminating the residual noise component from the input signal and providing the input signal; Second means (12,13,13) for adding a slowly varying negative signal to said input signal so as to provide an output signal substantially corresponding to the varying data component.
14; 33); third means (18; 32) coupled to the output portion (16; 35) of the first means for sensing the output signal; and the output signal becomes more negative. A fourth means (20; 31, 34) for rapidly adding a positive signal to the signal being processed in response to the negative output signal sensed by said third means. The system described above, comprising: 2. The residual noise suppression system according to claim 1, wherein said input signal includes a rectified audio signal and a low frequency noise component. 3. The residual noise suppression system according to claim 2, wherein said second means includes a resistor (33) connected between said input portion (11) and ground. 4. The second means comprises a current source (13), a capacitor (14) connected in series therewith, one of its inputs connected to said input portion (11), and said current source. 3. The system of claim 2 including a subtraction device (12) having a second input connected to a midpoint of said capacitor. 5. The apparatus according to claim 1, wherein said third means includes a first terminal connected to ground.
4. The residual noise suppression system of claim 3 including an operational amplifier (32) having a second input connected to receive said output signal and said output signal. 6. The third means includes a comparator (18) having an input connected to the output portion (16) of the first means and another input connected to ground. 4. A residual noise suppression system. 7. The operational amplifier (32), wherein the fourth means comprises:
A diode (34) connected between the output of the first means and the output part (35) of the first means, and a capacitor (31) connected in series to the input part (11) of the first means. The residual noise suppression system according to claim 5. 8. The fourth means is responsive to the third means and connected in parallel with the second means (12-14), the negative DC sensed by the third means. 7. The residual noise suppression system of claim 6, including gating means (20) for selectively and rapidly reducing the output signal. 9. A method for suppressing residual noise in processing an input signal having a slowly varying residual noise level component and a rapidly varying data component therein, wherein the residual noise level component is substantially reduced from the input signal. Applying a slowly varying negative signal to the input signal to provide an output signal substantially equivalent to a rapidly varying data component of the input signal; sensing the output signal; Rapidly adding a positive signal to the signal being processed when a negative output signal is sensed to prevent the output signal from becoming more negative. 10. The method of claim 9 wherein said positive signal is added to the signal being processed by selectively and rapidly reducing the level of a slowly varying negative signal applied to an input signal. 11. A method for suppressing residual noise in processing an input signal having a slowly varying residual noise level component and a rapidly varying data component therein, said method comprising substantially converting said residual noise level component from said input signal. Applying a slowly varying negative signal to the input signal to provide an output signal substantially equivalent to a rapidly varying data component of the input signal; sensing the output signal; negative Reducing the fluctuating negative signal to prevent the output signal from becoming more negative when the output signal is sensed. 12. A noise suppression system for receiving an input signal having a residual noise level component and a data component therein, canceling the noise level component, and providing an output signal substantially corresponding to the data component. A first means (13, 14) for providing a signal corresponding to the residual noise level component; and an input signal for providing an output signal corresponding to the data component of the input signal. Second means (12) for subtracting a signal corresponding to the residual noise component, and the first means for preventing the signal to be subtracted from exceeding the value of the residual noise level component of the input signal. And a third means (18, 20) responsive to the output signal and coupled to the first means for limiting the value of the applied signal. Stem. 13. The method of claim 1, wherein the input signal comprises a rectified audio signal, the data component comprises auditory intelligence,
Means comprises a capacitor (14) and a current source (13) connected in series with each other, said second means comprising in the circuit one of its inputs connected to the midpoint of said current source and said capacitor. And a subtraction device (12) having another input connected to receive the input signal, wherein the third means includes, in a circuit, the input of the input connected to the output of the subtraction device. A comparator (18) having one and another input connected to ground in the circuit, said third means further comprising gating means (20) controlled by the output of said comparator;
13. The noise suppression system of claim 12, wherein said gating means is connected across said capacitor to selectively discharge said capacitor when the output of said subtraction device becomes negative. 14. The subtracting device includes a summing amplifier (12), the comparator includes an operational amplifier (18), the gating means includes a gate in a circuit connected to an output of the comparator, and the capacitor (14). 14. The noise suppression system of claim 13, including a field effect transistor (20) having its source and drain terminals connecting thereto across. (I) receiving a multi-frequency input signal having a residual noise level component and a data component therein;
(Ii) a noise suppression system for providing, in different frequency bands, a plurality of output signals, each of the output signals substantially corresponding to a data component in the frequency band; Parallel channels (71-75); each of the channels for filtering the input signal to allow a limited frequency band portion of the input signal to pass through its own filtering means of the channel. Filter means (71a-75a)
The frequency band portion of each of the passed input signals of the channel is substantially different from the frequency band portion of the other passed input signal of the channel.
To 75b), rectifying means (71b to 75b) in each of the channels for rectifying a frequency band portion of the input signal passed through the filtering means in the channel, and an output signal in each of the channels. The respective elimination means (71) of the channels for eliminating residual noise level components from the various rectified input signals in the channels to substantially correspond to the data components of the input signals therein.
c to 75c), wherein each of the means for eliminating the residual noise level component comprises: first means (13, 14) for providing a signal corresponding to the value of the noise level component; Second means for subtracting a signal corresponding to the value of the noise level component from the input signal to provide an output signal corresponding to the data component of the input signal (12)
Responding to the output signal to limit the value of the signal provided by the first means so as to prevent the signal to be subtracted from exceeding the value of the residual noise level component of the input signal; and A noise suppression system comprising a third means (18, 20) coupled to the first means. 16. The input signal includes an audio signal,
16. The noise suppression system of claim 15, wherein the data component includes auditory intelligence. 17. The first means includes a capacitor (14) and a current source (13) connected in series with each other, and wherein the second means is connected to an intermediate point between the current source and the capacitor in a circuit. A subtraction device (12) having one of its inputs connected and another input connected to receive the input signal, the output of the subtraction device corresponding to an auditory intelligence portion of the input signal. Providing said output;
The third means includes a comparator (18) having one of its inputs connected to the output of the subtraction device in a circuit and the other input connected to ground in the circuit, wherein the third means comprises: And further comprising gating means (20) controlled by the output of said comparator, said gating means traversing said capacitor for selectively discharging said capacitor (14) when the output of said subtraction device goes negative. 17. The noise suppression system of claim 16 connected. 18. The subtraction device includes a summing amplifier (12), the comparator includes an operational amplifier (18), and the gating means traverses in the circuit a gate connected to the output of the comparator and the capacitor. 18. The noise suppression system of claim 17, including a field effect transistor (20) having its source and drain terminals connected thereto. 19. The residual noise level component includes a slowly varying signal, the data component includes a rapidly varying component, and each of the means for canceling the residual noise level component receives the input signal. And first means including a circuit having an input portion (11) for outputting the data component and an output portion (16; 35); substantially eliminating the residual noise level component from the input signal; Continuous to the input portion of the first means for applying a slowly varying negative signal to the input signal to provide an output signal substantially corresponding to the rapidly varying data component of the input signal; Second means (12-14; 33), and third means (18; 32) for sensing the output signal.
And fourth means responsive to sensing a negative value in the output signal for rapidly adding a positive signal to the signal being processed to prevent the output signal from becoming more negative. 16. The noise suppression system according to claim 15, comprising: (20; 31,34,). 20. The method according to claim 19, wherein the second means includes the input part.
20. The noise suppression system of claim 19, including a resistor (33) connected between the resistor and ground. 21. The second means comprises a current source (13), a capacitor (14) connected in series therewith, one of its inputs connected to said input portion (11), and said current source. 20. The residual noise suppression system of claim 19, further comprising a subtraction device (12) having a second input connected to a midpoint of the capacitor. 22. The third means includes an operational amplifier (32) having a first input connected to ground and a second input connected to receive the output signal. 20 noise suppression systems. 23. The third means comprising: the output section (16).
22. The noise suppression system of claim 21 including a comparator (18) having an input connected to and another input connected to ground. 24. The fourth means, wherein the operational amplifier (3
2) a diode (34) connected between the output of the first means and the output part (35) of the first means, and a capacitor (31) connected in series to the input part (11) of the first means. 23. The noise suppression system of claim 22, comprising: 25. The fourth means is responsive to the third means and connected in parallel with the second means (12-14), the negative output sensed by the third means. 24. The noise suppression system according to claim 23, including gating means (20) for selectively rapidly reducing the signal. 26. The rate at which said second means adds said slowly varying negative signal to said input signal, the rate at which said second means provides selective suppression of residual noise in said different channels. 20. The noise suppression system of claim 19, which is selectively different. 27. A method for suppressing residual noise in processing a composite input signal having a slowly varying residual noise level component and a rapidly varying data component therein, the method comprising: removing the residual noise level component from the composite signal. Applying a slowly varying signal to the composite input signal to substantially cancel and provide an output signal having a predetermined polarity and substantially corresponding to a rapidly varying data component of the input signal; the output signal Sensing the output signal and ensuring that the output signal maintains the predetermined polarity. 28. A method for receiving a composite input signal having a residual noise level component and a data component, canceling the noise level component, and generating a unipolar output signal having a predetermined polarity and substantially corresponding to the data component. A noise suppression system for providing a signal corresponding to the value of the residual noise level component; and a unit corresponding to the data component of the input signal. A second signal for subtracting a signal corresponding to the residual noise level component from the composite input signal to provide a polarity output signal.
Means (12) for responding to the unipolar output signal and third means (18, 20) for ensuring that the unipolar output signal maintains the predetermined polarity. The above-mentioned system, characterized in that: