Ambient noise reduction arrangements

Ambient noise reduction devices {AMBIENT NOISE REDUCTION ARRANGEMENTS}

The present invention relates to an apparatus for reducing or eliminating main-surface noise perceived by a listener using earphones. In the present application, the term "earphone" relates to a device incorporating a loudspeaker that is placed outside the listener's ear. For example, part of a "pad-on-ear" or "shell-on-ear" enclosure or part of an assembly such as a mobile phone, which is positioned close to the ear.

The earphone loudspeaker may be connected to a source of voice or other sound that must be distinguished from ambient noise, or the loudspeaker may be provided alone to reduce ambient noise, however, the present invention is compatible with personal music players and cellular phones. This invention is particularly applicable to earphones used with mobile electronic devices, such as.

Currently, some earphones are wired directly to the sound source via short leads and connectors, and some are connected to a local sound generating device such as a personal music player or cell phone via a wireless link such as a "Bluetooth" format. The present invention can be used with both wired and wireless formats.

Existing ambient noise cancellation systems for earphones are based on one or the other of two completely different principles: a "feedback" method and a "feedforward" method.

The feedback method is based on the use of a miniature microphone placed directly in front of the earphone loudspeaker, inside a cavity formed between the ear and the inside of the earphone shell. The signal from the microphone is connected back to the loudspeaker through a negative feedback loop (inverting amplifier), forming a simple servo system, where the loudspeaker takes a null sound pressure level from the microphone. We want to keep making Although this principle is simple, when trying to implement it, the problem is that the upper frequency of the operation is limited to about 1 kHz or less. Moreover, effective passive acoustic attenuation must be provided to prevent the ingress of ambient noise above this 1 kHz limit, and this is done by providing an earmuff-like seal designed to block these frequencies. An attempt to improve the performance of a recent feedback system is described in document US 2005/0249355 A1.

Also, if music or voice is to be provided to the user's earphones, a method must be provided to prevent these signals from being removed by the feedback system, and this process causes an undesirable spectral throw in the acoustic properties of the earphones. (spectral troughs) and peaks. Moreover, this type of feedback system requires that the operating cavity be substantially decoupled from its surroundings, and although "padded" feedback devices were proposed approximately two decades ago, earphones of this type are not yet commercially available. is the state Feedback systems are prone to "howl" around oscillations when switches are on or when operating conditions change.

The device according to the invention therefore uses the feedforward principle entirely, which is shown in its basic form in FIG. 1 .

In the feedforward operation, the microphone (A) is located outside the earphone shell (B) to detect the ambient noise signal. The signal detected by the microphone A is inverted at C and added to the drive signal and applied to the loudspeaker D, thus generating a "cancellation signal". It is intended to be adjacent to the earphone loudspeaker exit port within the cavity between the earphone shell (B) and the listener's outer ear (E), so that evanescent cancellation between the cancellation signal and the incoming ambient acoustic noise signal occurs. will be. For this to happen, the magnitude of the cancellation signal must be substantially equal to the magnitude of the incoming noise signal and the polarity must be opposite (ie, the phase must be inverted or shifted 180 degrees with respect to the noise signal).

The earphone shell (B) is typically a foam pad or similar device to provide a comfortable fit to the listener's outer ear (E) and/or to help reduce ambient noise reaching the listener's ear. has the

In principle, implementing feedforward ambient noise cancellation is straightforward. The basic operating system for use with conventional earphones can be inexpensively assembled using a simple electret microphone capsule and a pair of op amps so that the analog signal is amplified and mixed with the earphone audio drive signal before mixing. is reversed This is done through a gain-tunable device, such as a potentiometer, to make the magnitude of the cancellation signal equal to the magnitude of the ambient noise. Some measures of noise rejection can be achieved with the present invention, but may not be complete. Nevertheless, the feedforward principle forms the basis of many commercially available earphones today. However, even if the cancellation signal is optimally tuned and balanced, a significant amount of residual noise signal still remains, so in general, most commercially available systems are only required to operate at frequencies lower than about 1 kHz, So only a slightly larger bandwidth is provided than the bandwidth of the feedback method. Given that the voice spectrum extends to 3.4 kHz, any relevant denoising system requires bandwidths that exceed the capabilities of currently available systems to significantly improve the intelligibility of a conversation, for example over a telecommunications link. .

It is an object of the present invention to provide a device capable of reducing a significant amount of ambient noise up to at least 3 kHz.

According to the present invention, there is provided an apparatus for reducing ambient noise, the apparatus comprising a housing, supported within the housing, and disposed adjacent to an entry point of the auditory canal of a listener's ear, the sound energy loudspeaker means for directing to the listener's ear; means for converting ambient noise into an electrical signal, wherein the electrical signal is applied to the loudspeaker means to generate an acoustic signal opposing the ambient noise, the acoustic signal comprising the ambient noise at the entry position generated by the loudspeaker means substantially timed to the arrival of

The advantage obtained by this means is that the ambient noise arriving at the entry point takes advantage of the time difference between the detection of the ambient noise at the microphone means and the arrival of the ambient noise at the entry point to the listener's ear canal. and a noise reduction or elimination signal that is substantially time-aligned is generated.

In some preferred embodiments, an array of microphone means is provided around the ear pads forming part of the housing for the loudspeaker, the loudspeaker means having a known radial distance from the loudspeaker means to the respective microphone means. distance) within the housing. In another preferred embodiment, an array of microphone means is provided around and radially spaced from a loudspeaker aperture of a mobile telephone handset. In either case, the radial path followed by ambient noise from the microphone means to the vicinity of the loudspeaker is sufficient for the noise reducing acoustic signal to be generated in either case, as will be described in detail later herein. to ensure that the required temporal alignment is achieved.

In a particularly preferred embodiment, the relative position and arrangement of the microphone means and the loudspeaker means with respect to the incoming ambient noise is chosen to take account of the performance characteristics of the loudspeaker means to ensure the necessary temporal alignment.

It is particularly preferred for the microphone means to be positioned such that it responds substantially uniformly throughout to incident ambient sound from a significant angular range.

In some preferred embodiments, at least three, preferably at least five microphone means are provided for sensing incoming ambient noise. Moreover, when such a number of microphone means are provided, they are preferably arranged at substantially equal angles around a common locus.

Such a trajectory may conveniently carry elements of an electrical componentry configured to interconnect the microphone means and/or to carry the output to a common location for processing.

This electrical component may be provided as a printed circuit, and the processing may include combination, phase reversal, and amplitude adjustment.

Any or both of these microphone means may be exposed to ambient noise via apertures and conduits, which may be tuned to one or more selected ambient noise characteristics to provide enhanced noise reduction with respect to one or more specific characteristics. It may further include any acoustic elements that allow it to be provided.

Such acoustic elements as mentioned above may consist of Helmholz resonators and/or quarter-wave resonant conduit or may consist of Helmholz resonators and/or quarter-wave resonant conduits. conduits may be included.

In all such embodiments, it is preferred that the acoustic projection axis of the loudspeaker means is substantially aligned with the longitudinal axis of the listener's ear canal.

In order to clearly understand the present invention and to easily achieve the effects of the present invention, some embodiments of the present invention are described with reference to the drawings by way of example only.

1 shows, in basic form, the elements of a feedforward noise-reduction arrangement.

FIG. 2 schematically shows a feedforward system in the prior art of the kind shown in FIG. 1 with its associated acoustic path; FIG.

3 shows a curve representing the timing change due to a difference in the length of the acoustic path shown in FIG. 2 .

4 shows an apparatus for reducing noise according to an embodiment of the present invention.

Fig. 5 shows an acoustic path illustrating the operation of the embodiment of the present invention shown in Fig. 4;

6 schematically shows the acoustic leakage path around the earphone device.

7 shows a curve representing the timing change due to a difference in the length of the acoustic path shown in FIG. 5 .

8 is a curve showing the performance of a commercially available noise reduction earphone device.

9 is a curve compared with FIG. 8 and also a curve showing the performance of the embodiment of the present invention shown in FIG.

Fig. 10 shows an equivalent circuit for an electret microphone, and shows an operating characteristic curve thereof.

11 shows an integrated electret microphone array and buffer amplifier circuit.

12 shows an earphone of a device according to an embodiment of the present invention.

13 illustrates a device configured for use with a wireless earphone, in accordance with an example of the present invention.

14 illustrates an apparatus configured for use with a cellular phone, according to an example of the present invention.

15 is a three-dimensional diagram showing the sensitivity of the noise reduction effect to changes in amplitude and phase.

16 shows a curve showing the maximum possible noise reduction with different time delay errors.

Before the detailed description of the present invention, as a general description, referring to FIG. 2 , an important problem associated with using a conventional feedforward device of the kind described with respect to FIG. 1 is described. FIG. 2 uses the same reference numerals as reference encoding applied to the corresponding components in FIG. 1 .

Figure 2a shows a simple feedforward ambient noise cancellation system, wherein the microphone (A) is mounted on the earphone shell (B) in a central position, which is a simplified top view of a cross-section of a listener wearing an earphone through the ear canal plane. , and is shown in the front direction (azimuth angle 0 degrees) in the upper part of the figure.

When a sound wave SF is incident from the front direction, the wave-front arrives at the listener's eardrum G slightly later than it reaches the microphone A, because the sound This is because the path lengths are different as shown. After traveling to the microphone through a path of length X and also to a point P under the earphone which intersects the longitudinal axis of the ear canal H, where the point lies at the entry point of the ear canal, the wave is To reach the tympanic membrane (G), an additional distance Y must be advanced. Path length Y is approximately the sum of the length of the ear canal (H) (typically 22 mm), the depth of the outer ear (J) (typically 17 mm), and the small air gap above the ear of about 5 mm, for a total 44 mm, and the corresponding transit time is 128 μs.

However, as shown in Fig. 2b, if the direction of incidence is from a lateral position (i.e. 90 degrees azimuth), the wave front SL reaches the microphone A first, but the aforementioned entry position P and the additional path distance to the tympanic membrane (G) is much greater than before. Here, after traveling through path X to both the microphone itself and a parallel position in line with the rim of the earphone shell B, the wave SL travels an additional distance Z as well as Y before reaching the tympanic membrane G. have to proceed

Consequently, it can be considered that there is a significant difference in the relative arrival times of the wavefronts SF and SL at the microphone A and the point P (and consequently the tympanic membrane G), which Depending on the direction of the sound source, this difference in arrival time is due to the difference Z between the two paths.

These time-of-arrival variations can be measured using an "artificial head" system that mimics the auditory properties of the human head and ear, where a suitable ear canal simulator or equivalent can accurately propagate to the location of the eardrum. It can be integrated into the auditory structure to allow measurement of delay. For example, publication US 6,643,375 describes one possible measurement system developed by the inventor. Measurements are made by mounting a reference loudspeaker at a distance of about 1 meter from the artificial head, which has an earphone and microphone system, is in the same horizontal plane as the ear, is positioned at a selected azimuth, and is repeated at a frequency of 8 Hz. A fast transition wave, such as a ms rectangular pulse, is applied to the loudspeaker. This allows the arrival of wavefronts to be accurately identified by synchronously and simultaneously recording signals from (a) a microphone in the ear canal of the artificial head and (b) an externally mounted microphone on the earphone shell.

A typical pair of measurements from a centrally mounted ambient noise microphone (which is attached to a 50 mm diameter earphone module mounted on an artificial head (with canal) and ear system) is shown in Figure 3 in the form of two waveform pairs, Each pair is synchronously recorded simultaneously from the oscilloscope. Each pair of waveforms represents a signal from an artificial head microphone located at the next extraction position in the MC and an external ambient noise recording microphone in the ME. Fig. 3 shows that when there is a sound source in the anterior direction (0 degrees azimuth, for example SF in Fig. 2a), the sound wave front arrives at the external microphone at 161 μs before reaching the tympanic membrane. However, when the sound source is at 90 degrees in azimuth (eg, SL in Fig. 2b), the time difference is much larger, i.e. 300 μs, and in the intermediate direction, the time difference of arrival is between these two extreme values and , thus changing by about 140 μs.

Since the time difference of arrival varies greatly with the direction of the sound source, it is difficult to know which temporal alignment of any kind should be achieved for this type of configuration. Even if a system is made to operate in one particular direction, it may be ineffective in all other directions.

A further problem in implementing a simple feedforward device of the kind shown in Fig. 1 arises due to the finite response time nature of a typical loudspeaker, which is important with respect to the critical timing factors involved. was discovered by These issues are explained in more detail later.

Turning now to specific examples of the present invention, a configuration in accordance with some embodiments of the present invention, which will now be disclosed, uses a distributed microphone array, which with a feedforward system for earphone related ambient noise cancellation, an earphone shell, or It is formed around a casing or pad.

This configuration enables improved temporal alignment of the cancellation signal to the ambient noise signal in the tympanic membrane, which causes the two problems mentioned above with the conventional feedforward system (i.e., (a) the tympanic path length due to the change in the sound source direction). This is possible by adequately handling the significant variations in ambient noise with respect to (b) the time delay associated with the electroacoustic transducer). Consequently, the present invention provides a feedforward based configuration that operates up to a higher frequency than hitherto possible, which is also substantially omnidirectional in nature.

As a first step, a plurality of microphones are used to detect ambient noise, and these microphones are positioned to reduce variations in the sound path length with the sound front direction. In practice, the single microphone configuration used in the prior art is improved by using only two microphones, but preferably three or more microphones are used. In the description immediately following of a preferred embodiment of the present invention, an array of five uniformly distributed microphones is used, spaced at intervals of 72 degrees around the rim of the earphone.

Figure 4 is three simplified diagrams showing one basic embodiment of the present invention. Fig. 4A is a top view of the earphone 10 when it is in the outer ear, showing the radial cross-sectional axis A-A' through one of the five microphone positions, and Fig. 4B shows the axis A-A' It is the front elevation of the section through the . This embodiment of the present invention is intended to include an array of distributed five subminiature electret microphones 21 , 22 , 23 , 24 and 25 mounted in a housing close to the perimeter of the rim 20 of the earphone capsule 10 . is appearing Details of the microphone mounting configuration are shown in FIG. 4C . Each microphone, such as 21, is mounted such that its inlet port 26 is exposed to ambient air through a short conduit 27, which is typically approximately the same width as the microphone and 0.5 mm high; And the length is a few millimeters. These dimensions are not critical and the conduit is shown in plan in FIG. 4A . The back of each microphone is also exposed to the environment via a leak path (not shown) to balance the internal pressure across the microphone diaphram. Preferably each microphone, such as 21, is rigidly mounted on a common Printed-Circuit Board (PCB) 28 to simplify electrical lead-out connections, And it is desirable to configure the microphones in parallel to simplify the associated electronic circuitry as will be explained later. The microphone is acoustically isolated from the loudspeaker as far as possible. Preferably, the microphone inlet ports are aligned around the rim of the earphone, but if desired they may be aligned on the outermost surface.

The earphone capsule 10 includes a casing 11 which acts as a chassis for the various components, in which is mounted a high compliance microspeaker 12, typically 34 mm in diameter, the diaphragm of which is It is exposed through a protective grille 13 at the bottom edge, to which a foam pad 14 is attached to provide comfort to the listener's outer ear. Alternatively, for improved acoustic isolation at higher frequencies (>4 kHz), conventional foam-filled leather annular rings may be used instead of these. The loudspeaker is provided with a rear cavity 15 , typically having a volume of a few ml, in order to provide a highly compliant loading, and preferably this is the fundamental resonance of the loudspeaker 12 . To minimize , damping is performed using acoustic foam. Also preferably, the backside volume is peripherally exposed through one or more apertures, such as 16, to maximize backside loading compliance. These holes are spaced 10 mm or more from the microphone inlet ports, such as 26.

In padded earphones, the earphone units are acoustically impermeable, so each earphone assembly acts as an acoustic baffle in close contact with the listener's pinna. A thin foam-rubber pad 14, typically between 3 mm and 6 mm thick, covers the surface of the earphone to provide a comfortable surface for the listener and some small degree of acoustic sealing between the outer ear and the surroundings. used to cover The latter, although less effective at low frequencies below about 4 kHz, serves three purposes (i.e. (a) to increase the low-frequency response of the earphone, and (b) to limit outward-facing sound emission from the earphone to the surroundings. and (c) reducing the ingress of ambient noise from the surrounding environment).

Important features according to this embodiment of the present invention are as follows. Because the earphone 10 acts as an obstruction, the path of acoustic leakage from the periphery to the tympanic membrane must travel half the diameter of the earphone assembly before reaching an axial entry position for the ear canal. Therefore, by positioning the microphones 21 to 25 on the rim 20 of the earphone or close to the rim 20 of the earphone, the ambient noise signal can be obtained before reaching the eardrum and applied to the electroacoustic transducer 12 . Therefore, it is possible to compensate for the inherent response time of the electroacoustic transducer 12 . Moreover, this applies to wavefronts arriving from all directions.

For example, in a configuration such as that described with reference to FIG. 4 , FIG. 5 shows the acoustic path from the frontal noise source NF to the tympanic membrane at an azimuth of 0 degrees ( FIG. 5A ) (as in FIG. 2 ). , and the acoustic path from the lateral noise source NL to the tympanic membrane at an azimuth angle of 90 degrees (further 5b). The acoustic path is simplified and conceptually divided into three sections (X', Y' and Z') to illustrate this particularity.

In this step, only the signal path through one of the microphones 21 is considered for the sake of a clear initial description and for the purpose of approximating and quantifying the time delay involved.

Referring to FIG. 5A , the front source wave front NF first arrives at the rim 20 of the earphone, where it is detected by the microphone 21 , the path following the path X. The wave front (NF) must then go through the radius of the earphone to reach the eardrum (G), then through the depth of the outer ear (J), and then through the length of the ear canal (H) ( shown here as the combined path (Y')). However, the cancellation signal bypasses path Z'. Consequently, assuming that there is no time delay in the feedforward electronic circuit, the cancellation signal is the signal of the ambient noise at the entry point P on the central axis of the ear canal (i.e., the junction of the paths Y' and Z'). It can be sent to the earphone's loudspeaker before it arrives. By matching the time of flight of the radial path length Z' to the response time of the loudspeaker of the earphone, substantially accurate time alignment can be achieved. Conveniently, this can be realized in practice with the earphone diameter. For example, a 60 mm diameter earphone has a radial path distance of 30 mm, which corresponds to a flight time of 87 μs, which is well matched to the inherent response time of many small earphone loudspeakers.

Referring now to FIG. 5B , it can be seen that a similar process occurs for a noise wave fret NL arriving from a lateral source at 90 degrees in azimuth. The presence of the earphone 10 is preventing the wave NL from traveling directly to the ear, and thus must travel around the structure, following a path similar to that of the wave front NF from the front of FIG. 5A . After reaching the rim 20 of the earphone, the wave front NL is detected by one or more microphones, such as 21, and then, to reach the eardrum G, the radius (path of the earphone) (Z')), followed by the depth of the outer ear and the length of the ear canal (path (Y')). So, as before, the loudspeaker can be driven with the cancellation signal obtained from the rim microphones before the arrival of the noise signal at the entry position P on the central axis of the ear canal.

In the preceding description, in order to roughly quantify the time delay involved and to simplify the description, the effect of only a single microphone 21 was considered. However, it should be noted that the process becomes slightly more complex. The inventors observed that when the wave front reaches the earphone unit and then proceeds, a continuous diffraction process takes place under the rim of the earphone as shown in FIG. 6 , and the wave front completely passes through the earphone assembly. Until then, there is a subsequent acoustic leakage into the cavity between the outer ear and the earphone.

Figure 6 shows that this process takes place for the wave front NF' of frontal origin, where the interaction process is longest. 6a shows the arrival of the wave front NF of the leading edge (front edge) of the earphone casing 11 , where the leakage path L1 is under the earphone. When the wave NF' travels through the earphone and reaches an intermediate position (FIG. 6B), an ingress leakage path occurs through diffraction around and below the earphone rim 20. When the wave front NF completes travel in the earphone 10 and leaves the rear (rear) edge ( FIG. 6C ), the wave front diffracts around the earphone rim 20 and below and behind the earphone 10 , thus ) and the sound pressure level between the outer ear.

This phenomenon varies depending on the direction. If the wave front comes from the front noise source, then the acoustic energy is properly distributed in time, ie over the period the wave front spends to propagate the 60 mm earphone shell (this is about 175 μs). However, if the incoming wavefront is normally incident on the earphone (ie 90 degrees in azimuth), the energy arrives all at once and is not distributed properly in time.

Thus, the impulse response (and associated transfer function) from the periphery to the tympanic membrane varies greatly with the sound source direction, as already shown in FIG. 3 . The front impulse response has (a) a much smaller peak amplitude and (b) a longer duration than the lateral impulse response. However, the configurations according to the invention take this into account automatically, since effectively they integrate the sound pressure level around the rim of the earphone and generate a signal representing the overall dynamic leak-driving SPL as a function of time. .

A typical pair of measurements from a distributed array of 5 microphones (which is incorporated into a 50 mm diameter earphone module mounted on an artificial head (with canal) and ear system) is shown in Figure 7 in the form of two waveform pairs. and each pair is synchronously recorded simultaneously from the oscilloscope. Each pair of waveforms represents a signal (MC) from the artificial head microphone and a signal (MA) from an array of 5 microphones distributed. 7 shows that when the sound source is in the front direction (0 degree azimuth), the sound wave front reaches the external microphone 200 at 200 μs before reaching the eardrum. However, when the sound source is at 90 degrees in azimuth, this time difference is only slightly larger (ie, 250 μs). In the intermediate direction, the arrival time difference can be between the two values, thus changing only by a total of 50 μs (compared to a 140 μs change of a single microphone). This ±25 μs variation provides a temporal alignment that we can show as sufficient to achieve -10 dB rejection at 2 kHz.

It should be noted that the impulse response of ambient leakage-to-eardrum (Fig. 3) and that of ambient-to-microphone array (Fig. 7) are not directly comparable to each other. This is not the case, since the former involves the effect of sound path and movement under the earphone, whereas the latter does not. These figures are provided merely to illustrate the similarity of changes in magnitude and duration between the two and that they are similarly direction dependent.

Conceptually, the total ambient noise leakage to the earphone/outer ear cavity can be considered as the sum of many elemental radial leakage paths, with a central node centered on the longitudinal axis through the ear canal. Congregate at the entry point containing Thus, the ambient noise signal at the conceptual center of the radial elemental leakage paths is the time-dependent sum of the elemental contributions after propagation from the rim 20 of the earphone 10 to the position P.

If the elementary leakage paths have similar acoustic impedance, the ambient noise SPL at the conceptual center P of the radial elementary leakage paths, after the radial propagation delay, is around the rim 20 of the earphone. represents the time-dependent sum of each SPL at the outer points of the elementary leakage paths. This conceptual central ambient noise SPL is what drives the outer ear and ear canal, and it is this signal that the distributed ring-shaped microphone array 21-25 detects and registers prior to its occurrence, according to the principles of the present invention.

The effectiveness of the present invention is best achieved by comparing the performance of one of the best commercial over-the-ear noise canceling earphones with that of having a distributed array of five microphones of the kind shown in FIG. 4 according to an embodiment of the present invention. can be proven Commercial earphones from major manufacturers were chosen to perform the best of the four different sets evaluated. A five-microphone distributed array signal was used in a simple feedforward denoising configuration, without any filtering or other signal processing (other than amplification and inversion) to show this effect. Measurements are made on an artificial head system that features an artificial ear device with an ear canal. The earphones are placed on the artificial head, the noise cancellation is switched off, and the frequency response to the artificial head is in the horizontal plane, using standard methods (both MLS sequence and swept sinusoidal). Measured from a small loudspeaker at a distance of 1 meter and an azimuth of 45 degrees. Next, the feedforward denoising is switched on and the measurement is repeated. The results were processed to eliminate loudspeaker coloration by subtracting previous reference measurements made with a reference microphone (B&k 4003), and these are shown for commercially available earphones in FIG. 8 , , and FIG. 9 shows an earphone including a five-microphone array according to an embodiment of the present invention.

8 shows measurements obtained from commercially available over-ear noise canceling earphones in the form of three frequency response graphs obtained from ear canal microphones in an artificial head. The first graph (A) shows the response of the microphone to be pulled out without proper placement of the earphone to serve as a reference. The second graph (B) shows the response when the earphone is properly placed and noise cancellation is switched off, and the third graph (C) shows the response when noise cancellation is switched on.

The shape of the reference response (A) is caused by the resonant properties of the outer ear and ear canal, with its maximum peak at approximately 2.6 kHz. When the earphone is properly positioned (graph (B)), incoming ambient frequencies above 2 kHz are placed under passive attenuation by a foam cushion that partially seals the earphone to the outer ear as shown in Figure 1b. However, in the 400 Hz to 1.5 kHz range, the motion applied on the earphones causes the ambient noise level at the eardrum by +6 dB at 1 kHz, due to the currently existing cavity between each earphone and its respective outer ear. actually increases Graph (C) is the result when the noise canceling circuit is switched on. It can be seen from this that in the range from 300 Hz to the upper limit of 1.5 kHz, there is a slight decrease at most by only -6 dB. The reduction at 1 kHz is only -3 dB.

9 shows a similar and direct comparison set of responses for earphones comprising a distributed five-microphone array, conveniently a pad-like configuration rather than an over-ear configuration, resulting in greater acoustic leakage from the periphery to the ear; there will be When the earphones are properly positioned (graph (B)), the response above 4 kHz is reduced by passive attenuation, and the peak response is slightly increased by about 3 dB due to resonance as mentioned before. However, when noise rejection is switched on (graph (C)), the response decreases significantly in the 300 Hz to 3.5 kHz range, and decreases by about -10 dB in the 300 Hz to 1 kHz range. The reduction at 1 kHz is approximately -12 dB, and now the upper limit is 3.5 kHz. Table 1 below summarizes the improvements in noise cancellation provided by a distributed five microphone array compared to high quality commercially available earphones.

<img file="KR20080098385A_D0001.tif" />

Practically, in the arrangement according to the present invention, in terms of cost and complexity, there is a trade-off between the accuracy of the signal matching (between the cancellation signal and the noise signal) and the selection of the number of microphones. There is also a balance to be sought between the required signal "lead time" required from the microphone and the physical diameter of the earphone assembly, since it is the diameter of the distributed microphone that determines this lead time. The following description guides the practical implementation of the present invention in this respect.

To achieve accurate time alignment, the difference in arrival time between the surrounding microphone(s) and the ear canal microphone is equal (substantially similar to) the system response time from the electroacoustic transducer (i.e., the loudspeaker of the earphone) to the ear canal microphone. )must do it.

Keep in mind, since each acoustic path shares a common path element (shown as Y' in Fig. 5) to the outer ear and to the ear canal and tympanic membrane, the first solution is to choose an appropriate radius for the distributed circumferential array. , is to make the time delay associated with the remaining path elements (shown as Z' in Fig. 5) equal to the transducer response time.

The first step is to measure the response time of the electroacoustic transducer selected for the earphone driving module. If the transducer response time is, for example, 70 μs (a typical value), then this corresponds to an acoustic path length of about 24 mm, so that the acoustic center of the distributed microphone array has a diameter of approximately 48 mm or a diameter near it. It should be the center of the circle.

However, the acoustic path is not as straight and simple as this, and it is best to measure the difference in arrival times and adjust the radius accordingly to obtain the best accuracy. In practice, most transducers suitable for this purpose have response times in the range of 70 μs to 100 μs, so distributed microphone array diameters in the range of 40 mm to 60 mm are well suited for these values.

Next, the number of microphones to be used in the array must be selected. Of course, ideally a large number is better than a small number, because if too small a number is used, there can be a risk of some quantization effects. If we want to suggest a reasonable criterion that a temporal alignment better than 40 μs (corresponding propagation distance is about 14 mm) is desirable, we investigate the geometry of the wave passing through a circular microphone array with radius R, and the equation As, their angular separation (<img file="KR20080098385A_D0002.tif" />), it is possible to investigate a simple and approximate relationship with respect to the effective distance D through which a transverse wave between individual microphones passes.

<img file="KR20080098385A_D0003.tif" />

Through implantation, for a microphone-to-microphone time interval smaller than 40 μs (D = 14 mm), if R = 30 mm, <img file="KR20080098385A_D0004.tif" /> = ~60 degrees, so 6 microphones should be used. However, this is only a rough guideline.

When the earphones are placed in slightly different positions when they are worn on the listener's ear, the presence of acoustic leakage properties and considerable variability in different acoustic path lengths is unavoidable in this kind of system. This, along with the impact of any small design trade-offs already made, tends to limit the performance of the system, so the noise suppression properties are still characterized by "finite" suppression crossover points. However, this is generally well observed above 3 kHz, in contrast to the sub 1 kHz crossover frequencies measured in prior art devices.

Precise orientation of the individual microphones is important, but not essential. In order to best represent the SPL at the entrance to the leak path, the microphone entrance (eg 26 ) should be located close to the rim edge 20 adjacent the listener's head. For example, this ensures that the wave diffracted back at the rear edge of the earphone (FIG. 6) is correctly registered. If a microphone inlet port, such as 26, is placed away from the listener's head, it registers a propagating wavefront before diffraction around the earphone and at the back under the earphone, and at a much higher frequency than ambient noise reaching the eardrum. It contains a lot of energy, since the latter undergoes back diffraction.

When defining a microphone array, the most suitable transducer is a miniature electret microphone, as is well known to those skilled in the art. The present inventors used various sub-miniature electret microphones from various manufacturers, whose sizes ranged from 6 mm in diameter by 5 mm in length to 3 mm in diameter by 1.5 mm in length. The microphone should have a relatively flat frequency response (±3 dB between 200 Hz and 10 kHz), and the variation in sensitivity between microphones should be less than ±3 dB at 1 kHz. (These specifications are typical for the 3 mm diameter x 1.5 mm long microphone used by the inventor)

Regarding the electrical construction of the microphone array, each microphone includes an integral FET buffer amplifier, and thus the output impedance is only a few kΩ. Figure 10a shows a simple equivalent circuit of a typical microphone capsule, wherein the electret film is shown with a small capacitance (C1) of about 100 pF and a high parallel leakage resistance (R2), typically 100 MΩ, which is an n-channel JFET. It is connected between the gate of (Junction Field-Effect Transistor) (J1) and the ground. In use, the JFET drain connection is connected to a low voltage source (V1), typically +3V, through a load resistor (R1). The transmission characteristics of a typical JFET microphone capsule are, I<sb>D</sb>/V<sb>DS</sb> In the form of a characteristic, it is shown in Figure 10b. It can be seen that there is a saturation region where the drain-source voltage is greater than about +1 V, and the associated saturation current is about 250 μA. In this region, the conductivity of the JFET is V<sb>DS</sb>is largely independent of and is primarily controlled by the gate-source voltage difference (ie, audio-dependent changes in voltage across the electret (not included here for brevity). Figure 10a shows a typical load resistor of 6 kΩ, which combined with a +3 V bias voltage results in a device current of 250 μA, a V of 1.5 V.<sb>DS</sb><sb></sb>value, and causes a DC output voltage level of +1.5 V on the output node.

However, the microphone signal is relatively small (a few mV in magnitude) and therefore still needs amplification. Rather than using a separate preamplification stage for each microphone and then a voltage summing stage, it is convenient to construct a single amplification stage that is applied to all of the microphones simultaneously. One way to achieve this is to connect both microphones in parallel. However, what is important in this particular type of configuration is that both microphones are operated in their saturation region, otherwise intermodulation occurs, because a change in current in one microphone changes the common node voltage, which also This is because it changes the current in the microphone's integral FETs. For example, in Figure 10a, if four additional microphone capsules are simply connected in parallel with the original using a load resistance R1 of 6 kΩ, the output voltage V<sb>DS</sb>) is reduced to only 200 mV, and the current flowing through each microphone JFET is only 90 μA. It is under saturation, where V<sb>DS</sb>Any change in V will cause a large change in the device current, thus modulating the audio signal.

To avoid this intermodulation phenomenon, the selected microphone type I<sb>D</sb>/V<sb>DS</sb> The characteristics should be measured, as shown in Figure 10b, and then the saturation region and current of the microphone and the integral FET can be determined. This allows a single suitable bias resistor to be selected for the entire microphone array, allowing safe operation without intermodulation effects. For example, if it is necessary to form a 5 microphone array in parallel according to the above features, the load resistor R1 should be reduced substantially to 1.2 kΩ in this case. This results in the current flowing through each of the five devices to be a satisfactory saturation current of 250 μA (1.25 mA overall), and as before, V<sb>DS</sb> The value is 1.5 V.

Figure 11 shows a preferred circuit configuration in which five microphones are connected in parallel to a suitable buffer amplifier X1A (in this case the gain factor is 28). The output of this stage is fed to the loudspeaker drive stage through a gain adjustable stage for trimming the amplification and, if necessary, to the inverter following any polarity change in the subsequent circuit and speaker connections. As supplied, it can be used to drive a feedforward system of the general kind shown in FIG. 1 .

A simple basic embodiment of the invention has already been described with reference to Fig. 4, wherein a circular array of microphones is arranged around the rim of the earphone. Fig. 12 shows such a configuration mounted on the headband, reference numerals corresponding to those of Fig. 4 . A variant of these embodiments is to include the associated electronic components (power supply, preamplifier, inverter and audio driver) on an internal printed circuit board (PCB) integrated into the structure of the earphone casing. This is convenient for reducing external cabling, and reduces the cost of adding volume and weight to the overall device. This also provides the convenience that in manufacturing, for example, a microphone can be mounted directly around the edge of a circular PCB, in which the loudspeaker is electrically connected via spring contacts and thus "snap-together". configuration becomes possible. The acoustic partition may be held between the microphone and the rear volume of the loudspeaker using one or more suitable closed-cell foam polyurethane gaskets around the rim.

Another practical embodiment of the present invention is shown in FIG. 13 , in a wireless earphone (Bluetooth) configuration 30 . In this example, three microphones, each having an input port 31 , 32 , and 33 , are placed around a centrally placed loudspeaker (not shown, hidden by the outer surface 34 of the housing). are distributed The earphone 30 is also conventionally formed with an earclip 35 and a lip microphone boom 36 .

Fig. 14 shows another practical embodiment of the present invention, wherein a distributed microphone array is designed in a cellular phone handset unit 40 and is also designed in the form of a three microphone array. The individual input ports for the three microphones of the array are shown at 41, 42, and 43 respectively. A conventional microspeaker outlet port is shown at 44 .

In general, considering the starting point of the present invention from the conventional feedforward concept, as described with reference to FIG. And this is true, although many improvements have been made by using associated electronic filtering or by using adaptive filters to "tune out" periodic noise.

Previous proposals for feedforward configuration are based on the principle that both the incoming ambient noise signal and the signal driven through the earphone loudspeaker will be subjected to various deformations, for example by acoustic engineering within the earphone shell cavity. same. These modifications are considered to modify the amplitude response of the signal and prevent total rejection from occurring. However, there is no consideration due to the phase of the two signals, and it is proposed that if these various transfer functions are to be combined mathematically, an ideal electronic filter should be created to account for and predict both these effects.

It should be understood that, according to the present invention, the relative phase of the cancellation signal with respect to the ambient noise signal is considered to have at least equal importance to the relative amplitudes of the two signals.

Although various prior publications on ambient noise cancellation refer to the use of electronic filters to modify the amplitude response, there is no clear explanation regarding timing or phase response. For example, US 6,069,959 describes a complex filtering scheme for use with a feedforward denoising system, and discloses a number of graphs showing the amplitude response, but no description or reference to the timing or phase response. none.

There are some significant practical difficulties in implementing the above methods in measuring the various transfer functions and then combining them to form the necessary filter function.

The inventors of the present invention consider that the directionality of the foregoing transfer functions is important, and recognizes that this factor has not been observed before.

The inventors of the present invention also consider that it is not appropriate to use a transfer function obtained from a single angle measurement for use in a diffuse sound-field such as used on a daily basis.

In view of the weak results of several prior attempts to improve ambient noise cancellation systems, they resort to very sophisticated methods such as the use of adaptive filters. Paper summarizing the latest technology (Title: "Adaptive feedback active noise control headset: implementation, evaluation and its extensions", Authors: WS Gan, S Mitra and SM Kuo, IEEE Transactions on Consumer Electronics, 51, (3), August 2005 ) has been published. This method uses a digital signal processor (DSP) to analyze and identify various components of incoming noise, mainly for repetitive noises, and then modify the electronic filter in real time to optimize the Attempts to provide a removal signal. However, despite considerable mathematical and engineering efforts, these methods have had limited success. For example, the article (title: "Analogue active noise control", author: M Pawelczyk, Applied Acoustics, 63, (2002), pp. 1193-1213) contains an overview of the state-of-the-art in the art and have. From Fig. 15 of this paper, it can be seen that the cancellation bandwidth of the state-of-the-art adaptive system is limited to a frequency of about 500 Hz or less. Also, Pawelczyk says such a system cannot suppress impulsive non-repeatable noise.

Thus, it is clear that prior art publications are missing or neglecting the importance of the phase response of the cancellation signal to the incoming ambient noise signal. Moreover, the result of incorrectly matching the amplitudes of the two signals is not quantified.

To find out how sensitive the denoising process is to changes in amplitude and phase at the same time above and below the optimal value, we calculate the residual amount of (unremoved) noise (the "residual" signal) in a proportional (percent) form. Analyzes were performed to define the effectiveness of the noise removal process with respect to the noise level, and for the logarithmic reduction of the noise sound pressure level (SPL) in dB.

A slightly surprising result is that the tolerances are very tight, which is even necessary for a decent amount of noise rejection. If 65% rejection (-9 dB) is to be achieved (residual signal = 35%), then the amplitude of the rejection signal must match the amplitude of the noise signal within ±3 dB, and at the same time the phase of the signal is ±20 degrees ( 0.35 radians).

Fig. 15 shows a three-dimensional surface plotting the residual noise ratio as a function of amplitude and phase deviation from perfect matching, from which the relationship of critical properties becomes apparent. The &gt;50% rejection signal (-6 dB or better) is represented by a very narrow funnel-shaped lowest gray area that falls to the center of the bottom of the graph. Any deviation from this ideal region greatly reduces the effectiveness of the system.

The present invention provides an improved ambient noise cancellation configuration for earphone users, which is valid for frequencies up to 3 kHz and above, in contrast to the sub 1 kHz limit of currently available commercial products. Moreover, an advantage of the present invention is that it is comfortable in use and the amount of noise cancellation is electrically controllable, all of these features being highly desirable for use with mobile electronic devices.

In contrast to various conventional feedforward signal processing publications that emphasize only the amplitude response of a signal as a function of frequency, the present invention takes into account the critical importance of the relative phase of the signal.

Various conventional techniques involving signal processing based on various fixed transfer functions, each measured from a single selected spatial direction, where these are assumed to be valid for use in a diffuse sound field (omnidirectional). In contrast to the method, the arrangement according to the invention accommodates the change in the transfer function depending on the sound source direction, thereby providing an omnidirectional diffuse sound field noise reduction or noise reduction means.

The present invention is based on a novel principle that the cancellation signal must be aligned to be substantially "time aligned" with the incoming ambient noise signal at the listener's eardrum, and to ensure correct time alignment of the signals at the listener's eardrum. In addition, it provides a configuration that ensures direction-independent matching of the amplitudes of the two signals.

After the previously mentioned analysis performed by the present inventors on the sensitivity of the residual signal with respect to the amplitude and relative phase of the noise signal and the cancellation signal, the following conclusions were reached, that the exact phase relationship can be determined by electronic filtering or It cannot be obtained or adjusted by means of adaptive feedback or adaptive filtering, and the only means to achieve the correct phase relationship is to provide a "time-aligned" system. What this means is that the cancellation signal must be designed to be substantially time aligned with the incoming ambient noise signal.

However, this is not direct, since the ambient noise signal itself is acoustic and not electronic, and is therefore not available for correction using signal processing means.

16 contains two graphs plotting the residual noise level in dB as a function of frequency versus saturation, where the ambient noise and the rejection signal are equal in magnitude but in the time domain by only 80 μs and 40 μs, respectively. misaligned. The 80 μs period represents the time consumed by a sound wave traveling about 27 mm in air under a standard indoor environment. At low frequencies, up to 1 kHz, there is a modest amount of rejection (-6 dB), but this amount decreases as the noise frequency further increases, until a "crossover" point (here 2 kHz) is reached. This crossover frequency represents a point at which time-misalignment corresponds to one sixth of the period of the noise signal (π/3 radians). At these frequencies above the crossover point, in time misalignment, the cancellation signal is in-phase rather than out-of-phase with the noise signal, so that instead of the occurrence of vanishing wave cancellation, the cancellation signal is in-phase. occurs, and thus the resulting signal becomes larger than the original noise signal. The maximum point occurs when the time misalignment value equals one half of the wave period, with the residual signal being 6 dB greater than the original noise signal.

Currently, and as previously mentioned, various commercially available active noise cancellation systems are not effective above 1 kHz at best, and rely on passive attenuation by ear pads to reduce noise entry above 1 kHz. The second graph (solid line) of FIG. 16 shows that in order to achieve a noise rejection criterion of -6 dB at 2 kHz, the temporal alignment of the ambient noise and the rejection signal must be achieved with an accuracy of 40 μs or better. , which corresponds to a sound wave path length distance of only 14 mm in air. For a more practical noise rejection criterion of -10 dB at 2 kHz, the temporal alignment accuracy should be better (less than) 25 μs.

Although the aforementioned analysis is based on sinusoidal waveforms, it is clear that this is also directly applicable to random and non-repetitive waveforms, in that precise temporal alignment can result in total rejection of the noise signal.

As a result of ignoring the inherent time delay of electroacoustic transducers, problems also arise in conventional feedforward systems. Most earphone applications, in that the acceleration of the voice coil (and diaphragm) is proportional to the current flowing through the coil (depending on the applied voltage), and thus the sound pressure level (force per unit area) is directly proportional to it. It is assumed that the response time of the electroacoustic transducer used for

In practice, however, air connected to the diaphragm presents a complex acoustic load to the diaphragm in terms of acoustic inertia, acoustic mass, and acoustic resistance. This results in a finite response time that depends on many factors. From the inventor's experience, this is typically greater than 70 μs for microspeakers of very small diameter (16 mm), and typically 100 μs for earphone-type loudspeakers with a diameter of 38 mm.

The response time of a small loudspeaker can be measured by mounting the speaker on an obstruction plate, a reference grade microphone (B&amp;K type 4003) mounted coaxially to the speaker diaphragm, very close at a distance of about 2 mm be fitted As before, by driving the speaker with a rectangular waveform, the oscilloscope can be used to observe the microphone signal, and can be used to drive the signal synchronously and simultaneously, and to measure the rise time and response time of the speaker. can The propagation delay through a 2 mm separation distance is about 6 μs, and this can be subtracted from the measurements to yield the inherent loudspeaker response time. For one 34 mm loudspeaker used by the inventors, the measured response time is about 76 μs, and thus the intrinsic response time is about 70 μs, which corresponds to a sound wave path length distance of 24 mm.

This is in the sense that, if accurate time alignment is to be achieved, the cancellation signal must be delivered to the earphone loudspeaker millions of seconds before the microphone actually detects the signal, simply to compensate for the transducer delay. It creates an important conceptual problem for feedforward systems.

In general, the system response time is determined by (a) intrinsic loudspeaker response (described above) and (b) the outer ear edge from the loudspeaker diaphragm, then the depth of the concha cavity, and finally the ear to the microphone at the tympanic membrane position. It is the sum of the propagation times to the canal (path Y in Fig. 2). A typical response time is 247 μs.

Regarding the amplitude matching of the cancellation signal to the noise signal, the ambient noise signal travels through a complex acoustic path that appears between the earphone pad and the outer ear, the outer ear cavity and the auditory canal until it reaches the eardrum. and ends by the tympanic membrane. This network of conduits and cavities actually forms an acoustic filter that alters the spectral properties of the noise signal before reaching the eardrum. Both the frequency response and phase characteristics are varied as mentioned in the prior art. However, what the inventors have found is that since the earphone/outer-ear acoustic structure is common to both the ambient noise signal path to the tympanic membrane and the earphone loudspeaker to the tympanic membrane, the spectral modification that occurs to both signals is surprising. that is so similar. In fact, what the inventors have found is that if the microphone shows a reasonably flat frequency response, and the earphone loudspeaker also has a relatively flat frequency response, little or no pulse shaping is needed.

This observation is in contrast to some prior art publications in which signal processing based on various frequency domain transfer functions is used. Instead we use a time domain method.