GB2286945A - Noise reduction system - Google Patents

Abstract

A noise reduction system particularly but not exclusively for a vehicle, includes a first microphone 11 for receiving sound in the presence of a background noise and for generating a first audio signal, a second reference microphone 12 for receiving background noise and for generating a second audio signal, first signal treatment means T1 to treat the first audio signal and second treatment means T2 to treat the second audio signal, there being means 25 to sum output signals from the first and second treatment T1, T2 means whereby the noise component in the first audio signal generated by the first microphone 11 is at least partly cancelled. <IMAGE>

Description

Title: Noise Reduction System Description of the Invention This invention relates to a noise reduction system and is particularly concerned with a noise reduction system where it is desired to improve the clarity of speech or other sound received in the presence of background noise, by a microphone. The invention has been developed particularly but not exclusively for use in a vehicle such as an aircraft e.g. a helicopter, where background noise can particularly adversely affect the clarity of speech transmission in an audio system.

According to one aspect of the invention we provide a noise reduction system comprising a first microphone for receiving sound in the presence of background noise and for generating a first audio signal, a second reference microphone for receiving background noise and for generating a second audio signal, first treatment means to treat the first audio signal and second treatment means to treat the second audio signal and means to sum respective first and second output signals from the first and second treatment means whereby the noise component in the first audio signal can be at least part-cancelled.

The first and second signal treatment means may each comprise means to segment the respective audio signals into a plurality of frequency bands, there being filter means to filter the respective segmented audio signals in each frequency band.

At least one of the first and second signal treatment means may comprise means for adjusting the amplitude and phase of the respective signal being treated, so that the first and second output signals from the first and second treatment means are in phase and the amplitude of the first output signal attributable to the background noise is matched as closely as possible to the amplitude of the second output signal, to provide for as much cancellation of the noise component of the first treated signal as possible when the output signals are summed together. Of course, where the first and second audio signals generated by the microphone are generally in phase, it would be necessary to invert one of the signals, prior to or preferably subsequent to treatment, so that the background noise component in the first output signal from the treatment means is cancelled when the second output signal is summed with it.

According to a second aspect of the invention we provide a vehicle having installed therein a noise reduction system according to the first aspect of the invention. The vehicle may be an aircraft.

According to a third aspect of the invention we provide a method of reducing noise in an audio system which has a first microphone for receiving sound in the presence of background noise and for generating a first audio signal, the method comprising the steps of providing a second microphone for receiving background noise and generating a second audio signal, treating the first audio signal by a first treatment means and treating the second audio signal by a second treatment means, and summing first and second output signals from the respective first and second treatment means.

The method may include calibrating the noise reduction system by treating first and second audio signals generated in the absence of the sound the clarity of which is to be improved by the noise reduction system.

The invention will now be described by way of example only with reference to the accompanying drawings in which: Figure 1 is a schematic illustration of one embodiment of a noise reduction system according to the invention; Figure 2 is an illustration to explain part of a filter selection process performed by the system of Figure 1; Figure 3 is an illustration giving more detail of a signal treatment means of the embodiment of Figure 1; Figure 4 is a purely illustrative view of an aircraft in which is installed a noise reduction system in accordance with the invention.

Referring now to Figure 1 a noise reduction system 10 in accordance with the invention is incorporated into an audio system for speech transmission in a vehicle. The invention is particularly but not exclusively applicable to a vehicle comprising an aircraft as seen in Figure 4, but may be applicable in other situations where it is desired to improve the clarity of speech or other sound in the presence of background noise.

The system 10 includes a first microphone 11 which is located at the end of an antenna or stalk which is attached to a helmet 13 to be worn by an operator of the vehicle such as an aircrew member, and a second, reference, microphone 12 which is located preferably as close as possible to, but is shielded from, the first microphone 11. In the illustrated embodiment, the second, reference, microphone 12 is incorporated in the helmet 13 so that the second microphone 12 picks up as little as possible of the speech, but generates nearly the same signal due to the background noise, as the first microphone 11.

A first audio signal is generated by the first microphone 11 and is treated by a first signal treatment means T1 which includes various components which condition and treat the first audio signal and then provide, a first output signal. A second audio signal generated by the second microphone 12 is treated by a second treatment means T2 which includes various components which condition and treat the second audio signal and then provide a second output signal. The first and second output signals from the first and second treatment means are then summed as indicated at 25 in Figure 1, and as a result, that component of the first audio signal generated by the first microphone 11 due to background noise, is cancelled at least in part, thus to improve the clarity of speech which is then transmitted in the vehicle, or via a transmitter, from the vehicle.

An important feature of the system 10 is that audio signals generated by both microphones 11 and 12 are segmented, that is they are each separated into a plurality of frequency bands by a graphic equalizer which is generally indicated at 14.

The first and second treatment means each comprises a fourth order Butterworth type band pass filter bank 15 and 16 respectively which separates the individual signals generated by the microphones 11 and 12 into predetermined frequency bands.

Butterworth type filters were chosen for the filter banks 15 and 16 because of their smooth characteristic of amplitude and delay change with respect to frequency, which is an important consideration in audio work. Furthermore, Butterworth type filters permit wide band pass filters to be constructed readily from cascades of low pass and high pass filters.

The graphics equalizer 14 is controlled by a micro-controller 17 in response to peak amplitude detectors 18 effective in both of the filter banks 15 and 16. The micro-controller 17 includes algorithms for calculating signals 19 which are fed to an automatic amplitude control 20 which is effective to control the amplitude of both the first and second audio signals generated by the first and second microphones 11 and 12, upstream of the filter banks 15 and 16. The micro-controller 17 also provides signals 21 comprising amplitude weighting factors to the filter bank 16 as well as signals 22 comprising phase weighting factors, also to the filter bank 16.

The first output signal from filter bank 15 is fed to a signal inverter 24 before being summed at 25 with the output signal from the filter bank 16.

Alternatively, the second output signal from filter bank 16 could be inverted before being summed with the first output signal (non-inverted) from filter bank 15. Alternatively, where the microphones 11 and 12 generate audio signals which are generally 180 out of phase with one another, there may be no requirement for a signal inverter such as indicated at 24. In each case, when the first and second output signals are summed at 25, that component of the first audio signal attributable to background noise, will be at least partly cancelled and a cleaner signal as indicated at 26, in which the clarity of speech received by the microphone 11 is improved will result. The summed signal indicated at 26 may then be fed if desired to a low pass filter and treble/bass tone control device 27 for further speech enhancement. The low pass filter of control device 27 may be of the switched capacitance type or otherwise as desired. The treble/bass tone control 27 enables the clarity of the cleaned signal 26 to be further improved as desired.

Upstream of the treble/bass tone control device 27, in the present embodiment there is illustrated a switch 28 which is operated in response to a speech detector 29, which detects when the first microphone 11 is receiving speech, or is just receiving background noise. When the detector 29 detects that the first audio signal is attributable solely to background noise, the microphone 11 can then be muted. Also, the speech detector 29 is used to determine when calibration of the noise reduction system can be carried out, as hereinafter described.

Preferably, the first and second audio signals generated by the respective microphones 11 and 12, are treated upstream of the graphics equalizer 14, by a balancing device 30 which may emphasize high frequencies within the first and second audio signals which can then be more easily filtered from the audio signals. The device 30 may also be operable to equalize the signals generated by the microphones 11 and 12. Preferably the pre-emphasized and equalized first and second audio signals are then fed to individual bandwidth limiters 31 and 32 which may each comprise a fourth order Butterworth bandwidth limiter, which are operable to filter out frequencies within the first and second audio signals which are outside the usual frequency range attributable to speech.

Preferably, in order to calibrate the system 10 and otherwise to set parameters in it, the micro-controller 17 may be temporarily connected to a personal computer or other monitoring device as indicated at 33. The computer 33 may thus monitor system parameters, adjust system software parameters, and be used to store data accumulated in the system e.g. to hard disk, as well as to monitor a fault code store indicated at 34 which stores data relating to any fault identified by or in the micro-controller 17.

It is known that a frequency range of between 200 Hz and 6.1 KHz is essential for intelligible speech. In a particular application of the system 10, it was known that the vehicle in which the system was installed generated operational harmonics at 17.5 Hz and 35 Hz. Since these frequencies are well below that required for intelligible speech, the bandwidth limiters indicated at 31 and 32 which pretreat the first and second audio signals prior to the graphics equalizer 14, were in trials arranged to have a low frequency cut-off of 85 Hz.

At the other end of the range, it was known in a particular application, that considerable high frequency background noise of more than 6 KHz was generated by the vehicle and accordingly the bandwidth limiters 31 and 32 were arranged to have a high frequency cut-off of 6.91 KHz. Hence, the first and second audio signals which passed through the graphics equalizer 14, had a frequency in the range 85 Hz to 6.91 KHz only. However this still represents a wide frequency range and so it was decided to segment the speech frequency spectrum into four frequency bands for individual noise suppression treatment.

Illustrated in Figure 2, are the four separated frequency bands of one of the filter banks 15 and 16. It was found that by arranging corner or cut-off frequencies as follows, speech clarity was considerably improved when the signals were reconstituted in each of the filter banks 15 and 16, and also when the output signals were summed. The five corner or cut-off frequencies indicated in Figure 2 are: f0 = 85 Hz; fl = 256 Hz; f2 = 768 Hz; f3 = 2.3 KHz; 4 = 6.91 KHz.

The corner or cut-off frequencies were deliberately chosen to be coincident between the low pass and high pass edges of adjacent frequency bands, and a simple three times multiple to ensure simple synthesis via divide-by-three from a higher clock frequency. Thus the first and second audio signals are sampled at 230.4 KHz which yields a corner or cut-off frequency of 2.3 KHz, and 76.8 KHz yielding a corner or cut-off frequency of 768 Hz and 25.6 KHz yielding a corner or cut-off frequency of 256 Hz.

Sampling frequency clock inputs to the filter bands 15, 16 are indicated at 23a, 23b respectively.

Referring now also to Figure 3 in which like reference numerals indicate similar parts to those indicated in Figure 1, the extreme corner or cut-off frequencies of 85 Hz and 6.91 KHz include the frequency range 200 Hz to 6.1 KHz usually required for intelligible speech. Because anti-aliasing must be provided for the switched capacitance filters in filter banks 15 and 16, these two cut-off frequencies are most easily achieved by choosing bandwidth limiters 31 and 32 which provide for analogue/continuous filtering of the first and second audio signals.

Each of the filters of the filter banks 15 and 16 are preferably of the switched capacitance type. Subsequent to the first output signal from the filter bank 15 passing through the signal inverter 24, where provided, a fixed delay allpass filter 35 is preferably provided. Similarly, an all-pass filter 36 is preferably provided through which the second output signal from filter bank 16 passes.

However all-pass filter 36 is preferably of the programmable kind, thus to permit signal delay balancing between the output signals from filter banks 15 and 16, under adaptive control by the micro-controller 17 via weighting factor signals 22.

Thus in a noise reduction system of the illustrated embodiment of this invention, the first and second audio signals from the microphones 11 and 12 are each processed in real time by the two analogue filter banks 15 and 16. This hybrid system thus relies on a pair of fast analogue signal treatment channels to perform the actual noise cancellation, with there being a very slow digital loop with a 10 Hz sample rate to perform the automatic amplitude control achieved by the automatic amplitude control 20, and the amplitude and delay control as indicated by the signals 21 and 22 in Figures 1 and 3.

The noise reduction system contains four major elements namely: 1. noise reduction algorithms (in hardware, i.e. continuous); 2. adaptive minimisation algorithms (digital, i.e. in software); 3. built-in test (BIT) - digital; 4. special-to-type test equipment (STlE)e.g. computer 33, to enable: a) monitoring of system parameters in real time, b) adjustment of system parameters, c) data dump to store (e.g. hard disk) for later analysis, d) access to the fault code store 34.

Inevitably, conduction and absorption of noise by the helmet 13, the operator's skull and body will result in different noise signals being generated for the two microphones 11 and 12 in the absence of speech. The noise reduction system, by virtue of weighting factors 21 and the amplitude control 20, attempts to balance or equalize amplitude differences in the individual first and second audio signals, and to adjust any signal delay in either of the audio signals for minimum noise, during a calibration period, when there is no speech input from the microphone 11. A state of no speech input is recognised by speech detector 29.

The cancellation of the background noise picked up by the first microphone 11 requires both amplitude and delay matching of the audio signals from both of the microphones 11 and 12 after processing, for all frequencies treated in the filter banks 15 and 16. The micro-controller 17 ensures that both treated signals are amplitude and delay matched through the use of frequencylocked, switched capacitance filters in the band pass regions in filter banks 15 and 16. Furthermore, the micro-controller 17 attempts to manipulate any noise delay differences in real time that may exist between the first and second audio signals generated by the microphones 11, 12 e.g. due to the distance between the microphones 11 and 12. The arrangement described permits any phase difference in the output signals from the filter banks 15 and 16 to be accommodated by advancing or retarding the output signal from the filter bank 16 to bring that signal as close as possible into phase with the output signal from the filter bank 15.

Reliance on the active, real time phase correction by the microcontroller 17 can be reduced by ensuring that the microphones 11 and 12 are optimumly installed. It was determined in tests that the microphones 11 and 12 should be located as close as possible together for passive phase shift minimization in the absence of noise reduction system balancing. Calculations show that, at 1.5 KHz, a microphone separation of one inch (2.5cm) results in a phase shift of 0.1 of a wavelength, or 360, and also that less than a 600 phase shift between the two microphones 11 and 12 is necessary at any frequency, to prevent the treated first and second audio signals being out of phase to such an extent that by summing the signals, the background noise would be amplified rather than cancelled.

Anti-aliasing is ensured by the switched capacitor arrays of filter banks 15 and 16 being clocked at one hundred times their corner or cut-off frequency, the lowest being 256 Hz with a clock frequency of 25.6 KHz.

The system shown in Figure 1 includes preferred input signal processing at 30 providing peak clipping of speech, pre-emphasis and microphone equalization prior to the 85 HZ to 6.91 KHz limiting provided by the bandwidth limiters 31 and 32.

Pre-emphasis is used in the noise reduction system of this invention to mask transmission noise at higher audio frequencies. Pre-emphasis precedes peak clipping of 12 to 20 dB if the speech-to RMS noise ratio is less than 15 dB. Peak clipping is implemented by employing an automatic amplitude control to normalise the nominal speech plus noise content of each of the signals in the filter banks 15 and 16, and for excessive noise to be amplitude limited by permitting transient saturation of each filter bank 15, 16.

The noise reduction system incorporates calibration limitation requiring the micro-controller 17 to detect the presence of speech by a signal from the speech detector 29. Only in the absence of speech does the micro controller 17 attempt to balance or equalize the signals from the two microphones 11 and 12 and effectively freezes all system settings except the automatic amplitude control 20 when speech is detected. An operator speaking continuously for example could defeat the ability of the noise reduction system to track ambient noise changes which normally require periods of operator silence during which to auto-calibrate and cancel noise. The operator's microphone mute switch 37 is routed through micro-controller 17 so that it can take advantage of periods of operator silence time adaptively to re-configure the noise reduction system.

Preferably eight bit multiplying digital to analogue converters are used for automatic amplitude control at 20 as indicated in Figure 3, and for introducing the weighting factors 21 and 22 to filter bank 16.

Various modifications may be made without departing from the scope of the invention. For example the amplitude and delay weighting factors 21 and 22 signals can be applied to the audio signal from the first microphone 11 in filter bank 15 instead of to the audio signal from the second, reference, microphone 12 in filter bank 16. Depending on the application, the respective signals from the microphones 11 and 12 can be separated into any desired number of frequency bands other than the four described in the illustrated embodiment, and the frequency ranges of the respective bands can also be varied.

Whereas switched capacitance four channel fourth order Butterworth type filter banks have been described, other suitable forms of filters for the signal treatment means T1, T2 could be used.

Whereas in the embodiment described, the microphones 11 and 12 are separated from one another, they could be arranged coincidentally, provided that they are effectively screened from one another so that the second, reference microphone 12 picks up as little as possible of the speech.

Alternatively or additionally, the first microphone 11 could be mounted in a mask worn by an operator with the second microphone 12 in the mask, or in the helmet 13 or other headgear.

The features disclosed in the foregoing description, or the following claims, or the accompanying drawings, expressed in their specific forms or in terms of a means for performing the disclosed function, or a method or process for attaining the disclosed result, as appropriate, may, separately or in any combination of such features, be utilised for realising the invention in diverse forms thereof.

Claims (17)

1. A noise reduction system comprising a first microphone for receiving sound in the presence of background noise and for generating a first audio signal, a second reference microphone for receiving background noise and for generating a second audio signal, first treatment means to treat the first audio signal and second treatment means to treat the second audio signal and means to sum respective first and second output signals from the first and second treatment means.

2. A system according to claim 1 wherein the first and second signal treatment means each comprises means to segment the respective audio signals into a plurality of frequency bands, and filter means to filter the respective segmented signals.

3. A system according to claim 1 or claim 2 wherein at least one of the first and second signal treatment means comprises means for adjusting the amplitude and phase of the respective signal being treated.

4. A system according to any one of claims 1 to 3 wherein the second microphone is shielded from the first microphone so as to pick up minimal sound.

5. A system according to any one of claims 1 to 4 wherein the first and second microphones are provided by headgear worn by an operator, the first microphone being mounted by one of a mask worn by the operator and a stalk attached to the remainder of the headgear, and the second microphone is mounted by the headgear close to the first microphone.

6. A system according to claim 5 wherein the headgear is a helmet and the second microphone is built into the helmet.

7. A noise reduction system substantially as hereinbefore described with reference to and as shown in the accompanying drawings.

8. A vehicle having installed therein a noise reduction system according to any one of claims 1 to 7.

9. A vehicle according to claim 8 which is an aircraft.

10. A vehicle substantially as hereinbefore described with reference to and as shown in the accompanying drawings.

11. A method of reducing noise in an audio system which has a first microphone for receiving sound in the presence of background noise and for generating a first audio signal comprises the steps of providing a second microphone for receiving background noise and generating a second audio signal, treating the first audio signal by a first signal treatment means and treating the second audio signal by a second signal treatment means, and summing first and second output signals from the respective first and second treatment means.

12. A method according to claim 11 which includes segmenting the respective first and second audio signals each into a plurality of frequency bands and filtering the respective segmented signals.

13. A method according to claim 11 or claim 12 which includes adjusting the amplitude and phase of at least one of the respective first and second audio signals being treated.

14. A method according to any one of claims 11 to 13 wherein the first and second audio signals are generally in phase, one of the first and second signal treatment means including means to invert the respective signal being treated.

15. A method according to any one of claims 11 to 14 which includes the step of calibration, by treating first and second audio signals generated by the first and second microphones, in the absence of sound which the first microphone is intended normally to pick up.

16. A method of reducing noise in an audio system substantially as hereinbefore described with reference to and as shown in the accompanying drawings.

17. Any novel feature or novel combination of features described herein and/or in the accompanying drawings.