WO2025094006A1 - Regulation of noise reduction - Google Patents

Abstract

Presented herein are techniques for auto-regulating the strength of a noise reduction process/system, such as a Deep Neural Network (DNN)-based noise reduction (DNR) process, based on an estimate of a signal-to-noise ratio (SNR) of an input signal that is determined based on a DNN output (noise reduced signal). More specifically, in accordance with embodiments presented herein, the DNR strength is set based on an SNR that is determined/estimated from a ratio of an input signal to a noise reduced signal (e.g., determined from an DNN output).

Description

Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 REGULATION OF NOISE REDUCTION BACKGROUND Field of the Invention [0001] The present invention relates generally to techniques for regulating noise reduction. Related Art [0002] Medical devices have provided a wide range of therapeutic benefits to recipients over recent decades. Medical devices can include internal or implantable components/devices, external or wearable components/devices, or combinations thereof (e.g., a device having an external component communicating with an implantable component). Medical devices, such as traditional hearing aids, partially or fully-implantable hearing prostheses (e.g., bone conduction devices, mechanical stimulators, cochlear implants, etc.), pacemakers, defibrillators, functional electrical stimulation devices, and other medical devices, have been successful in performing lifesaving and/or lifestyle enhancement functions and/or recipient monitoring for a number of years. [0003] The types of medical devices and the ranges of functions performed thereby have increased over the years. For example, many medical devices, sometimes referred to as “implantable medical devices,” now often include one or more instruments, apparatus, sensors, processors, controllers or other functional mechanical or electrical components that are permanently or temporarily implanted in a recipient. These functional devices are typically used to diagnose, prevent, monitor, treat, or manage a disease/injury or symptom thereof, or to investigate, replace or modify the anatomy or a physiological process. Many of these functional devices utilize power and/or data received from external devices that are part of, or operate in conjunction with, implantable components. SUMMARY [0004] In one aspect, a method is provided. The method comprises: receiving at least one input signal at one or more input devices; processing the at least one input signal with a noise reduction process to generate a spectral gain for application to the at least one input signal; and determining a signal-to-noise ratio (SNR) of the at least one input signal based on the spectral gain. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 [0005] In another aspect, one or more non-transitory computer readable storage media are provided. The one or more -transitory computer readable storage media comprise instructions that, when executed by one or more processors of a hearing device, cause the one or more processors to: calculate a noise-reduction gain for application to one or more sound signals received at sound inputs of the hearing device; determine a signal-to-noise ratio (SNR) of the one or more sound signals based on the noise-reduction gain; apply the noise-reduction gain to the one or more sound signals to generate one or more noise-reduced signals; and mix the one or more noise-reduced signals with the one or more sound signals in accordance with a mixing ratio, wherein the mixing ratio is a function of the SNR. [0006] In another aspect, one or more non-transitory computer readable storage media are provided. The one or more -transitory computer readable storage media comprise instructions that, when executed by one or more processors of a hearing device, cause the one or more processors to: calculate a noise-reduction gain for application to one or more sound signals received at sound inputs of the hearing device; determine a signal-to-noise ratio (SNR) of the one or more sound signals based on the noise-reduction gain; determine a noise reduction strength parameter as a function of the SNR; and generate an output signal from the noise- reduction gain and the noise reduction strength parameter. [0007] In another aspect, a device is provided. The device comprises: one or more input devices configured to receive a noisy input signal, wherein the noisy input signal includes a target signal embedded in additive noise; a Deep Neural Network (DNN)-based noise reduction module configured to generate a DNN output signal from the noisy input signal; a signal-to- noise ratio (SNR) estimation module configured to estimate an SNR of the noisy input signal based on the DNN output signal; a mapping module configured to generate a noise reduction strength function that relates the estimated SNR to a noise reduction strength; and an application module configured to use the noise reduction strength function to generate an output signal representing the target signal.

Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 BRIEF DESCRIPTION OF THE DRAWINGS [0008] Embodiments of the present invention are described herein in conjunction with the accompanying drawings, in which: [0009] FIG. 1A is a schematic diagram illustrating a cochlear implant system with which aspects of the techniques presented herein can be implemented; [0010] FIG. 1B is a side view of a recipient wearing a sound processing unit of the cochlear implant system of FIG.1A; [0011] FIG.1C is a schematic view of components of the cochlear implant system of FIG.1A; [0012] FIG.1D is a block diagram of the cochlear implant system of FIG.1A; [0013] FIG. 2 is a functional block diagram of a noise reduction system, in accordance with certain embodiments presented herein; [0014] FIG.3 is a functional block diagram of another noise reduction system, in accordance with certain embodiments presented herein; [0015] FIG. 4 is a graph illustrating an example mapping function relating a priori SNR estimate to noise reduction strength (mixing ratio, alpha), in accordance with certain embodiments presented herein; [0016] FIG. 5 is a graph illustrating predicted SNR relative to true SNR, in accordance with certain embodiments presented herein; and [0017] FIG. 6 is a flowchart of an example method, in accordance with certain embodiments presented herein. DETAILED DESCRIPTION [0018] Noise reduction techniques, such as Deep Neural Network (DNN)-based noise reduction, sometimes referred to herein as DNR, have demonstrated good improvements in speech understanding and sound quality for hearing device users listening in noisy environments. Referring specifically to DNR the performance depends on several things, such as network complexity and size/quality of its training, and typically varies with input signal- to-noise ratio (SNR), where performance at high SNRs is typically better than performance at low SNRs. At low SNRs, where it is more challenging for algorithms to correctly separate out the clean target, the resulting heavy distortion of the input signal is undesirable. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 [0019] In order to optimize DNR benefit, the original input signal can be mixed with the noise reduced signal (generated based on a Deep Neural Network (DNN) output), where the mixing ratio is used to determine the overall strength of noise reduction in the final output signal. As such, it is useful to vary the strength of the noise reduction with SNR, for example using a higher DNR strength at high SNRs and a lower DNR strength at low SNRs. Presented herein are techniques for auto-regulating the strength of a DNR process based on an estimate of the SNR that is determined based on the DNN output (noise reduced signal). More specifically, in accordance with embodiments presented herein, the DNR strength is set based on an SNR that is determined/estimated from a ratio of an input signal to a noise reduced signal (e.g., determined from an DNN output). [0020] For ease of illustration, the techniques presented herein will be primarily described with reference to Deep Neural Network (DNN)-based noise reduction techniques/processes. However, it is to be appreciated that the techniques presented herein can be used with other types of noise reduction techniques/processes. That is, reference to Deep Neural Network (DNN)-based noise reduction techniques is illustrative. [0021] There are a number of different types of devices in/with which embodiments of the present invention may be implemented. Merely for ease of description, the techniques presented herein are primarily described with reference to a specific device in the form of a cochlear implant system. However, it is to be appreciated that the techniques presented herein may also be partially or fully implemented by any of a number of different types of devices, including consumer electronic device (e.g., mobile phones), wearable devices (e.g., smartwatches), hearing devices, implantable medical devices, wearable devices, etc. consumer electronic devices, wearable devices (e.g., smart watches, etc.), etc. As used herein, the term “hearing device” is to be broadly construed as any device that delivers sound signals to a user in any form, including in the form of acoustical stimulation, mechanical stimulation, electrical stimulation, etc. As such, a hearing device can be a device for use by a hearing-impaired person (e.g., hearing aids, middle ear auditory prostheses, bone conduction devices, direct acoustic stimulators, electro-acoustic hearing prostheses, auditory brainstem stimulators, bimodal hearing prostheses, bilateral hearing prostheses, dedicated tinnitus therapy devices, tinnitus therapy device systems. combinations or variations thereof, etc.) or a device for use by a person with normal hearing (e.g., consumer devices that provide audio streaming, consumer headphones, earphones, and other listening devices). In other examples, the techniques presented herein can be implemented by, or used in conjunction with, various implantable Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 medical devices, such as vestibular devices (e.g., vestibular implants), visual devices (i.e., bionic eyes), sensors, pacemakers, drug delivery systems, defibrillators, functional electrical stimulation devices, catheters, seizure devices (e.g., devices for monitoring and/or treating epileptic events), sleep apnea devices, electroporation devices, etc. [0022] FIGs.1A-1D illustrates an example cochlear implant system 102 with which aspects of the techniques presented herein can be implemented. The cochlear implant system 102 comprises an external component 104 that is configured to be directly or indirectly attached to the body of the user, and an internal/implantable component 112 that is configured to be implanted in or worn on the head of the user. In the examples of FIGs.1A-1D, the implantable component 112 is sometimes referred to as a “cochlear implant.” FIG. 1A illustrates the cochlear implant 112 implanted in the head 154 of a user, while FIG.1B is a schematic drawing of the external component 104 worn on the head 154 of the user. FIG.1C is another schematic view of the cochlear implant system 102, while FIG. 1D illustrates further details of the cochlear implant system 102. For ease of description, FIGs.1A-1D will generally be described together. [0023] In the examples of FIGs. 1A-1D, the external component 104 comprises a sound processing unit 106, an external coil 108, and generally, a magnet fixed relative to the external coil 108. The cochlear implant 112 includes an implantable coil 114, an implant body 134, and an elongate stimulating assembly 116 configured to be implanted in the user’s cochlea. In one example, the sound processing unit 106 is an off-the-ear (OTE) sound processing unit, sometimes referred to herein as an OTE component, that is configured to send data and power to the implantable component 112. In general, an OTE sound processing unit is a component having a generally cylindrically shaped housing 111 and which is configured to be magnetically coupled to the user’s head 154 (e.g., includes an integrated external magnet 150 configured to be magnetically coupled to an internal/implantable magnet 152 in the implantable component 112). The OTE sound processing unit 106 also includes an integrated external (headpiece) coil 108 (the external coil 108) that is configured to be inductively coupled to the implantable coil 114. [0024] It is to be appreciated that the OTE sound processing unit 106 is merely illustrative of the external devices that could operate with implantable component 112. For example, in alternative examples, the external component 104 may comprise a behind-the-ear (BTE) sound processing unit configured to be attached to, and worn adjacent to, the recipient’s ear. In general, a BTE sound processing unit comprises a housing that is shaped to be worn on the Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 outer ear of the user and is connected to the separate external coil assembly via a cable, where the external coil assembly is configured to be magnetically and inductively coupled to the implantable coil 114. It is also to be appreciated that alternative external components could be located in the user’s ear canal, worn on the body, etc. [0025] Although the cochlear implant system 102 includes the sound processing unit 106 and the cochlear implant 112, as described below, the cochlear implant 112 can operate independently from the sound processing unit 106, for at least a period, to stimulate the user. For example, the cochlear implant 112 can operate in a first general mode, sometimes referred to as an “external hearing mode,” in which the sound processing unit 106 captures sound signals which are then used as the basis for delivering stimulation signals to the user. The cochlear implant 112 can also operate in a second general mode, sometimes referred as an “invisible hearing” mode, in which the sound processing unit 106 is unable to provide sound signals to the cochlear implant 112 (e.g., the sound processing unit 106 is not present, the sound processing unit 106 is powered-off, the sound processing unit 106 is malfunctioning, etc.). As such, in the invisible hearing mode, the cochlear implant 112 captures sound signals itself via implantable sound sensors and then uses those sound signals as the basis for delivering stimulation signals to the user. Further details regarding operation of the cochlear implant 112 in the external hearing mode are provided below, followed by details regarding operation of the cochlear implant 112 in the invisible hearing mode. It is to be appreciated that reference to the external hearing mode and the invisible hearing mode is merely illustrative and that the cochlear implant 112 could also operate in alternative modes. [0026] In FIGs.1A and 1C, the cochlear implant system 102 is shown with an external device 110, configured to implement aspects of the techniques presented. The external device 110, which is shown in greater detail in FIG.1E, is a computing device, such as a personal computer (e.g., laptop, desktop, tablet), a mobile phone (e.g., smartphone), remote control unit, etc. The external device 110 and the cochlear implant system 102 (e.g., sound processing unit 106 or the cochlear implant 112) wirelessly communicate via a bi-directional communication link 126. The bi-directional communication link 126 may comprise, for example, a short-range communication, such as Bluetooth link, Bluetooth Low Energy (BLE) link, a proprietary link, etc. [0027] Returning to the example of FIGs.1A-1D, the sound processing unit 106 of the external component 104 also comprises one or more input devices configured to capture and/or receive input signals (e.g., sound or data signals) at the sound processing unit 106. The one or more Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 input devices include, for example, one or more sound input devices 118 (e.g., one or more external microphones, audio input ports, telecoils, etc.), one or more auxiliary input devices 128 (e.g., audio ports, such as a Direct Audio Input (DAI), data ports, such as a Universal Serial Bus (USB) port, cable port, etc.), and a short-range wireless transmitter/receiver (wireless transceiver) 120 (e.g., for communication with the external device 110), each located in, on or near the sound processing unit 106. However, it is to be appreciated that one or more input devices may include additional types of input devices and/or less input devices (e.g., the short- range wireless transceiver 120 and/or one or more auxiliary input devices 128 could be omitted). [0028] The sound processing unit 106 also comprises the external coil 108, a charging coil 130, a closely-coupled radio frequency transmitter/receiver (RF transceiver) 122, at least one rechargeable battery 132, and an external sound processing module 124. The external sound processing module 124 can be configured to perform a number of operations. For ease of illustration, only operations that are relevant to the techniques presented herein are represented in FIG. 1D, while other operations have been omitted. In particular, shown in FIG. 1D is a Deep Neural Network-based noise reduction (DNR) module 131, an SNR estimation module 133, and an SNR mapping module 135, and a mixing module 137. Again, the modules 131, 133, 135, and 137, are only intended to represent a subset of the functions/operations that can be performed by the external sound processing module 124. [0029] Each of the DNN module 131, the SNR estimation module 133, the SNR mapping module 135, and the mixing module 137 can be formed by one or more processors (e.g., one or more Digital Signal Processors (DSPs), one or more uC cores, etc.), firmware, software, etc. arranged to perform operations described herein. That is, the modules 131, 133, 135, and 137 can each be implemented as firmware elements, partially or fully implemented with digital logic gates in one or more application-specific integrated circuits (ASICs), partially or fully in software, etc. Although FIG.1D illustrates the DNN module 131, the SNR estimation module 133, the SNR mapping module 135, and the mixing module 137 as being implemented/performed at the external sound processing module 124, it is to be appreciated that these elements (e.g., functional operations) could also or alternatively be implemented/performed as part of the implantable sound processing module 158, as part of the external device 110, etc. The optional presence of these modules in the implantable sound processing module 158 (which itself could optionally be present or omitted in different arrangements) is represented by the dashed boxes in FIG.1D. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 [0030] Returning to the example of FIGs. 1A-1D, the implantable component 112 comprises an implant body (main module) 134, a lead region 136, and the intra-cochlear stimulating assembly 116, all configured to be implanted under the skin (tissue) 115 of the user. The implant body 134 generally comprises a hermetically-sealed housing 138 that includes, in certain examples, at least one power source 125 (e.g., one or more batteries, one or more capacitors, etc.) 125, in which RF interface circuitry 140 and a stimulator unit 142 are disposed. The implant body 134 also includes the internal/implantable coil 114 that is generally external to the housing 138, but which is connected to the RF interface circuitry 140 via a hermetic feedthrough (not shown in FIG.1D). [0031] As noted, stimulating assembly 116 is configured to be at least partially implanted in the user’s cochlea. Stimulating assembly 116 includes a plurality of longitudinally spaced intra-cochlear electrical stimulating contacts (electrodes) 144 that collectively form a contact array (electrode array) 146 for delivery of electrical stimulation (current) to the recipient’s cochlea. Stimulating assembly 116 extends through an opening in the recipient’s cochlea (e.g., cochleostomy, the round window, etc.) and has a proximal end connected to stimulator unit 142 via lead region 136 and a hermetic feedthrough (not shown in FIG.1D). Lead region 136 includes a plurality of conductors (wires) that electrically couple the electrodes 144 to the stimulator unit 142. The implantable component 112 also includes an electrode outside of the cochlea, sometimes referred to as the extra-cochlear electrode (ECE) 139. [0032] As noted, the cochlear implant system 102 includes the external coil 108 and the implantable coil 114. The external magnet 150 is fixed relative to the external coil 108 and the internal/implantable magnet 152 is fixed relative to the implantable coil 114. The external magnet 150 and the internal/implantable magnet 152 fixed relative to the external coil 108 and the internal/implantable coil 114, respectively, facilitate the operational alignment of the external coil 108 with the implantable coil 114. This operational alignment of the coils enables the external component 104 to transmit data and power to the implantable component 112 via a closely-coupled wireless link 148 formed between the external coil 108 with the implantable coil 114. In certain examples, the closely-coupled wireless link 148 is a radio frequency (RF) link. However, various other types of energy transfer, such as infrared (IR), electromagnetic, capacitive, and inductive transfer, may be used to transfer the power and/or data from an external component to an implantable component and, as such, FIG. 1D illustrates only one example arrangement. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 [0033] As noted above, sound processing unit 106 includes the external sound processing module 124. The external sound processing module 124 is configured to process the received input audio signals (received at one or more of the input devices, such as sound input devices 118 and/or auxiliary input devices 128), and convert the received input audio signals into output control signals for use in stimulating a first ear of a recipient or user (i.e., the external sound processing module 124 is configured to perform sound processing on input signals received at the sound processing unit 106). Stated differently, the one or more processors (e.g., processing element(s) implementing firmware, software, etc.) in the external sound processing module 124 are configured to execute sound processing logic in memory to convert the received input audio signals into output control signals (stimulation signals) that represent electrical stimulation for delivery to the recipient. [0034] As noted, FIG. 1D illustrates an embodiment in which the external sound processing module 124 in the sound processing unit 106 generates the output control signals. In an alternative embodiment, the sound processing unit 106 can send less processed information (e.g., audio data) to the implantable component 112 and the sound processing operations (e.g., conversion of input sounds to output control signals 156) can be performed by a processor within the implantable component 112. [0035] In FIG. 1D, according to an example embodiment, output control signals (stimulation signals) are provided to the RF transceiver 122, which transcutaneously transfers the output control signals (e.g., in an encoded manner) to the implantable component 112 via external coil 108 and implantable coil 114. That is, the output control signals (stimulation signals) are received at the RF interface circuitry 140 via implantable coil 114 and provided to the stimulator unit 142. The stimulator unit 142 is configured to utilize the output control signals to generate electrical stimulation signals (e.g., current signals) for delivery to the user’s cochlea via one or more of the stimulating contacts (electrodes) 144. In this way, cochlear implant system 102 electrically stimulates the user’s auditory nerve cells, bypassing absent or defective hair cells that normally transduce acoustic vibrations into neural activity, in a manner that causes the recipient to perceive one or more components of the input audio signals (the received sound signals). [0036] As detailed above, in the external hearing mode the cochlear implant 112 receives processed sound signals from the sound processing unit 106. However, in the invisible hearing mode, the cochlear implant 112 is configured to capture and process sound signals for use in electrically stimulating the user’s auditory nerve cells. In particular, as shown in FIG.1D, an Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 example embodiment of the cochlear implant 112 can include a plurality of implantable sound sensors 165(1), 165(2) that collectively form a sensor array 160, and an implantable sound processing module 158. Similar to the external sound processing module 124, the implantable sound processing module 158 may comprise, for example, one or more processors and a memory device (memory) that includes sound processing logic. The memory device may comprise any one or more of: Non-Volatile Memory (NVM), Ferroelectric Random Access Memory (FRAM), read only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical, or other physical/tangible memory storage devices. The one or more processors are, for example, microprocessors or microcontrollers that execute instructions for the sound processing logic stored in memory device. [0037] In the invisible hearing mode, the implantable sound sensors 165(1), 165(2) of the sensor array 160 are configured to detect/capture input sound signals 166 (e.g., acoustic sound signals, vibrations, etc.), which are provided to the implantable sound processing module 158. The implantable sound processing module 158 is configured to convert received input sound signals 166 (received at one or more of the implantable sound sensors 165(1), 165(2)) into output control signals 156 for use in stimulating the first ear of a recipient or user (i.e., the implantable sound processing module 158 is configured to perform sound processing operations). Stated differently, the one or more processors (e.g., processing element(s) implementing firmware, software, etc.) in implantable sound processing module 158 are configured to execute sound processing logic in memory to convert the received input sound signals 166 into output control signals 156 that are provided to the stimulator unit 142. The stimulator unit 142 is configured to utilize the output control signals 156 to generate electrical stimulation signals (e.g., current signals) for delivery to the user’s cochlea, thereby bypassing the absent or defective hair cells that normally transduce acoustic vibrations into neural activity. [0038] It is to be appreciated that the above description of the so-called external hearing mode and the so-called invisible hearing mode are merely illustrative and that the cochlear implant system 102 could operate differently in different embodiments. For example, in one alternative implementation of the external hearing mode, the cochlear implant 112 could use signals captured by the sound input devices 118 and the implantable sound sensors 165(1), 165(2) of sensor array 160 in generating stimulation signals for delivery to the user. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 [0039] As noted, Deep Neural Network (DNN)-based noise reduction (DNR) is one type of noise reduction technique that has demonstrated good improvements in speech understanding and sound quality for hearing device users listening in noisy environments. However, the performance of DNR depends on several things, such as network complexity and size/quality of its training, and can vary based on the with input signal-to-noise ratio (SNR) (e.g., the DNR typically performs well with signals having relatively high SNRs, but relatively poorly with signals having relatively low SNRs). In particular, with signals having relatively low SNRs, it is more challenging to correctly separate out the target signal (e.g., speech), which results in undesirable heavy distortion of the input signal. [0040] In order to optimize DNR benefit, the original input signal is mixed with the noise reduced signal (generated based on the DNN output), where the mixing ratio is used to determine the overall strength of noise reduction in the final output signal. As such, it is useful to vary the strength of the noise reduction (e.g., vary or control the mixing ratio) based on the SNR. For example, the mixing ratio can result in a higher DNR strength at high SNRs and a lower DNR strength at low SNRs. Presented herein are techniques in which the strength of the noise reduction (the DNR strength) to be auto-regulated based an estimate of the SNR that is determined based on the DNN output. More specifically, in accordance with embodiments presented herein, the DNR strength is a function of an SNR that is determined/estimated from a ratio of an input signal to a noise reduced signal. [0041] FIG. 2 is a functional block diagram illustrating an example Deep Neural Network (DNN)-based noise reduction (DNR) system 262, in accordance with certain embodiments presented herein. In particular, as shown in FIG. 2, the system 262 comprises a Deep Neural Network (DNN) noise reduction module 264, sometimes referred to herein as a DNN module, a signal-noise-ratio (SNR) estimation module 266, an SNR mapping module 268, and a mixing module 270. Operation of these various functional blocks, along with other aspects of system 262, are described below. [0042] As shown in FIG. 2, system 262 receives a noisy input, x(t), which is defined as the clean speech, s(t), embedded in additive noise, d(t), where t is the time index. This is represented as shown below in Equation 1. ^^^ ^^^ ൌ ^^^ ^^^ ^ ^^^ ^^^) (Equation 1) [0043] At filter block 261, a Fourier transform is applied to generate a frequency domain representation, ^^_^^ ^^^, of the noisy input (x(t)), using overlapping window frames with frame Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 n and frequency index k. The frequency domain representation is given as shown below in Equation 2. ^_^^ ^{^} _^^ ^{^} _{ൌ ^^^} ^{^} _^^ ^{^} _{^ ^^^^ ^^^ (Equation 2)} [0044] As shown, ^^_^ ^{^} ^^^{^}

of the input signal) is provided to the DNN module 264. The DNN module 264 calculates, from ^^_^^ ^^^, a DNN spectral gain, H_n(k) (e.g., DNN output signal or noise-reduction gain), which is applied to ^^_^^ ^^^ at block 263, which in turn generates a clean/noise-reduced output

^^^. That is, the noise- reduced signal (^ ^^_^^ ^^^) is created by application of the DNN spectral gain (Hn(k)) to the noisy input signal ( ^^_^^ ^^^^. This is given as shown below in Equation 3. ^^_^^ ^^^ ൌ ^^_^^ ^^^ ^^_^^ ^^^ (Equation 3)

[0045] As shown in FIG.2, Zn(k), is created by mixing the clean/noise- reduced signal (^ ^^_^^ ^^^) and the original noisy input ( ^^_^^ ^^^) in accordance with a mixing ratio or noise reduction strength parameter, alpha (α), which can vary over time and frequency, between the values of 0 and 1. This mixing process is represented by Equation 4, below. ^_^^ ^{^} _^^ ^{^} _{ൌ ^^^} ^{^} _^^ ^{^} _^^^ ^{^} _^^ ^{^} _{^ ൫1 െ ^^^} ^{^} _^^ ^{^} _{൯ ^^^} ^{^} _^^ ^{^} _{(Equation 4)}

[0046] And finally, in this specific example, a time domain output signal Zn(t) is obtained via an inverse FFT of ^^_^^ ^^^ In other examples, such as in the context of a cochlear implant, the inverse FFT is not needed and the frequency domain output ( ^^_^^ ^^^) is used for subsequent processing operations. [0047] In accordance with the techniques presented herein, the noise reduction strength parameter or mixing parameter, αn(k), is defined as a function of the a priori SNR ( ^^) which needs to be estimated by the algorithm, represented as shown below in Equation 5. ^^ _^^^ ^^^ ൌ ^^^ ^^^ (Equation 5) [0048] That is, the noise reduction strength parameter/mixing ratio varies with the SNR of the input signal, but the variation is non-linear (e.g., varies over time and frequency as a function of the SNR). To derive an estimate of the instantaneous a priori SNR ( ^^), its definition is used as a starting point, defined as shown below in Equation 6. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 ^^_^^ ^^^ ൌ ^{ா^|ௌ^^^^|మ^} ா_^|^^^^^|మ^ (Equation 6) [0049] In Equ 6, E{.} is the

the DNN output |Y| is used as an estimate of the speech power |S|², because it is assumed to be completely cleaned by the DNN and dominated by target speech, and can be directly observed after application of the DNN gains. This is shown below in Equation 7. ห ^^{^}^_^^ ^^^_ห ^ଶ ൌ | ^^_^^ ^^^|^ଶ (Equation 7) [0050] The noise power |D|² is estimated from the difference between the noisy input (X) and the clean speech estimate (Y), assuming speech (S) and noise (D) sources are uncorrelated, so cross power spectrum S.D is zero, as shown in Equation 8. ห ^^{^}^_^^ ^^^_ห ^ଶ ൌ | ^^_^^ ^^^|^ଶ െ | ^^_^^ ^^^|^ଶ (Equation 8)

[0051] Substituting to obtain an estimate of the a priori SNR based on the input (X) and output (Y) of the DNN is shown below in Equation 9. ^^{^} _^ ^{మ ^} _^ ^{|^^^^^|} ^ _^ ^{^} _{ൌ ^ ^^ ^^^^ 9)} [0052] In other words, in the

based on the original noise input signal (X) and the noise-reduced signal (Y) output by the DNN module 264 and, more specifically, based on a ratio of the original noise input signal (X) and the noise- reduced signal (Y). Stated differently, the SNR estimation is a function of the ratio of the original noise input signal (X) to the noise-reduced signal (Y). In the example of FIG. 2, the final output (Z) is determined from the original noise input signal (X) and the noise-reduced signal (Y) output by the DNN module 264, and noise reduction strength parameter/mixing ratio (alpha). [0053] As noted above, in the example of FIG.2, the noise-reduced signal (Y) is generated at 263 and used to generate the estimated SNR. FIG. 3 is a functional block diagram illustrates an alternative arrangement in which the noise-reduced signal (Y) is not generated and the Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 estimated SNR is estimated directly from the output spectral gain (H) (noise-reduction gain) of the DNN module 264. [0054] More specifically, shown in FIG.3 is a functional block diagram illustrating an example Deep Neural Network (DNN)-based noise reduction (DNR) system 362, in accordance with certain embodiments presented herein. In particular, as shown in FIG. 3, the system 362 comprises a Deep Neural Network (DNN) noise reduction module 364, sometimes referred to herein as a DNN module, a signal-noise-ratio (SNR) estimation module 366, an SNR mapping module 368, and a mixing module 370. Operation of these various functional blocks, along with other aspects of system 362, are described below. [0055] As shown in FIG.3, the noise reduction system 362 receives a noisy input, x(t), which defined as the clean speech, s(t), embedded in additive noise, d(t), where t is the time index (Equation 1, above). At filter block 361, a Fourier transform is applied to generate a frequency domain representation, ^^_^^ ^^^, of the noisy input (x(t)), using overlapping window frames with frame n and frequency index k . The frequency domain representation is given as shown in Equation 2, above. [0056] As shown, ^^_^ ^{^} ^^^{^} (frequency domain representation of the input signal) is provided to the DNN module 364. The DNN module 364 calculates, from ^^_^^ ^^^, a DNN spectral gain, H_n(k). As shown, the final output signal, Z_n(k), is created by applying the DNN spectral gain, Hn(k) to the original noisy input ( ^^_^^ ^^^, and then mixing the resulting noise-reduced signal with the original noisy input in accordance with a mixing ratio or noise reduction strength parameter, alpha (α), which can vary over time and frequency, between the values of 0 and 1. By substituting Y=H.X into Equation 4, above, it is possible to express Z in terms of H, X, and alpha, without the need for signal Y. This is shown below in Equation 10. ^^_^^ ^^^ ൌ ^^_^^ ^^^ ^^_^^ ^^^ ^^_^^ ^^^ ^ ൫1 െ ^^_^^ ^^^൯ ^^_^^ ^^^

(Equation 10) [0057] And finally, in this specific example, a time domain output signal Zn(t) is obtained via an inverse FFT of ^^_^^ ^^^ In other examples, such as in the context of a cochlear implant the inverse FFT is not needed and the frequency domain output ( ^^_^^ ^^^) is used for subsequent processing operations. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 [0058] In accordance with the example of FIG.3, substituting Y = H.X, H = Y/X into Equation 9, the a priori SNR estimate can be obtained directly from the DNN gain (H) without the need for the noise-reduced signal (Y). This is shown below in Equation 11. ^ ¹   ^^{^^^ ^^^ ൌ 1} ₁ (Equation 11)

[0059] In other words, in the example of FIG. 3, the SNR of the signal is determined directly from the output gain (H) of the DNN module 364. However, the output gain (H) represents a ratio of the original noise input signal (X) and a noise-reduced signal (Y). Stated differently, in FIG. 3, the SNR estimation is still a function of the ratio of the original noise input signal (X) to the noise-reduced signal (Y), represented as the output gain (H) of the DNN module. In the example of FIG. 3, the final output (Z) is determined from the output gain (H) and noise reduction strength parameter/mixing ratio (alpha). [0060] In summary, FIGs.2 and 3 show the noise reduction techniques presented herein in two configurations that are mathematically equivalent, where in FIG. 2 the signal Y is generated, while FIG.3 does not utilize the explicit calculation of signal Y. Instead, in FIG.3, the SNR estimate is based only on the output gain (H) of the DNN module. [0061] As noted above, the estimated SNR, whether determined as described above with reference to FIG.2 or FIG.3, is used to control the noise reduction strength parameter/mixing ratio α_n(k) (e.g., α_n(k) is defined as a function of the a priori SNR ( ^^)). FIG. 4 is a graph illustrating an example mapping function 472 relating the a priori SNR estimate to the noise reduction strength (e.g., illustrating how the noise reduction strength parameter/mixing ratio changes relative to the SNR). The noise reduction strength parameter/mixing ratio (alpha (α)) can vary over time and frequency, between the values of 0 and 1, and is set based on the SNR of the input signal. It is to be appreciated that the mapping function 472 shown in FIG. 4 is merely illustrative and that mapping functions can have different shapes for different recipients. [0062] In certain examples, the SNR estimate can be smoothed over an appropriate time period, commensurate with a time course suitable for making changes to the mixing ratio. Optional Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 integration over frequency, using a frequency importance scale such as the band importance function defined in ANSI S3.5–1997. American National Standard Methods for the Calculation of the Speech Intelligibility Index. New York: ANSI, 1997, can also be used. Smoothing of the SNR estimate is described below, but smoothing may be similarly applied to other signals in the chain, for example the noisy input (X), the clean speech (Y), the SNR, the noise reduction strength parameter/mixing ratio, etc. [0063] As noted, the SNR estimate can be smoothed over time, for example by using box car average by taking the mean over the last T time frames, as shown below in Equation 12. ^^{^} _^^ ^̅ _{^ ^^^ ൌ} ^{^} ் _∑ ^் ^_ୀ ^ି ^^{^} _^ ^{^} _{^^ି^^ ^^^ (Equation 12)} [0064] Smoothing can also

order IIR exponential smoothing using the following difference equation with forgetting factor β, as shown below in Equation 13. ^^{^} _{^^^ ^^^ ൌ ^^ ^} ^{^} _{^^^ ^^^ ^ ^1 െ ^^^ ^} ^{^} _{^^ି^^ ^^^ (Equation 13)}

[0065] The SNR can over bins, according to speech intelligibility importance function (SII), which provides a relative weight for each frequency bin, normalized so the sum of all weights equals zero, ∑^{^} ^_ୀ^ ^^ ^^ ^^^ ^^^ ൌ 1 , as shown below in Equation 14. ^^{^} _{^^ ൌ ∑} ^{^} ^_{ୀ^ ^} ^{^} _{^^^ ^^^. ^^ ^^ ^^^ ^^^ (Equation 14)}

[0066] Error! Reference source not found. 5 illustrates an example of the estimated SNR, using smoothing configured for box-car average over 5s, and SII(k) = 1/K (providing equal contribution from each frequency bin). Error! Reference source not found. shows the relationship between a priori SNR (True SNR) and its estimate (Predicted SNR) over a wide database of speech in noise samples (168 TIMIT sentences at each SNR, using a variety of noise types including babble, restaurant, café, city noise. etc., sourced from YouTube). [0067] In certain examples, the techniques presented herein utilize user interaction with the system for the purpose of designing and fine-tuning the noise reduction strength parameter/mixing ratio (e.g., adjusting the mapping function that maps a priori SNR estimate to the noise reduction strength) in order to suit individual requirements. In other words, a user can control, at least to some extent, the shape of the mapping function between the priori SNR estimate to the strength parameter (alpha). For example, the system could receive one or more Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 user input signals adjusting a relationship between the mixing ratio and the SNR, and the SNR mapping module could determine the mixing ratio based on the SNR and based on one or more user input signals. Such a system would allow the user to take control of the mixing ratio value (alpha) through an appropriate user interface, such as a slider or dial interface on a smart phone application. When under manual override, the function itself would be updated with the user selected noise reduction strength parameter/mixing ratio (alpha) value, according to the estimated SNR at the time. This optional user control of the noise reduction strength parameter/mixing ratio (e.g., receipt of user input signals adjusting a relationship between the mixing ratio and the SNR) is shown in FIGs. 2 and 3 by dashed arrows 275 and 375, respectively. [0068] In certain embodiments, a noise reduction system in accordance with embodiments presented herein could use different mapping functions for different use cases. For example, different mapping functions could be used in different noise types, different sound environments, different implementations of DNR, or for different end users with unique needs and/or preferences. More generally, the techniques presented herein could use metrics, other than SNR, that are calculated and used to at least partially control the strength of the noise reduction strength parameter. Examples include things like Root Mean Square (RMS) signal level, reverberation characteristics, and other signal statistics/attributes. [0069] FIG. 6 is a flowchart of an example method 680, in accordance with certain embodiments presented herein. Method 680 begins at 682 where at least one input signal (e.g., one or more sound signals) at one or more input devices. At 684, the at least one input signal is processed with a Deep Neural Network (DNN)-based noise reduction process to generate a spectral gain for application to the at least one input signal. At 686, a signal-to-noise ratio (SNR) of the at least one input signal is determined based on the spectral gain (the output of the DNN-based noise reduction process). [0070] In certain aspects, the spectral gain is applied to the at least one input signal to generate at least one noise-reduced signal. In addition, a noise reduction strength parameter is generated as a function of the SNR, and the at least one noise-reduced signal is mixed with the at least one input signal in accordance with the noise reduction strength parameter to generate at least one output signal. The at least one output signal can be used for subsequent processing operations. For example, in certain embodiments, the at least one input signal is a sound signal and the at least one output signal is an output signal used for subsequent processing and in delivering a hearing percept to a user. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 [0071] In accordance with certain embodiments of FIG.6, the spectral gain is applied to the at least one input signal to generate at least one noise-reduced signal, and the SNR of the at least one input signal is determined as a function of a ratio of the at least one input signal to the at least one noise-reduced signal. In accordance with certain embodiments of FIG.6, the SNR of the at least one input signal is determined directly from the spectral gain (e.g., without first generating the at least one noise-reduced signal). In such embodiments, the spectral gain represents/relates to a function of the ratio of the at least one input signal to the at least one noise-reduced signal, thus the SNR of the at least one input signal is still a function of a ratio of the at least one input signal to the at least one noise-reduced signal. [0072] In certain embodiments of FIG. 6, the SNR is an instantaneous SNR estimated based on the spectral gain. In other embodiments, the SNR is a smoothed SNR estimated over a period of time and/or estimated over a frequency range. [0073] In certain embodiments of FIG. 6, the noise reduction strength parameter is a function of the SNR and one or more user input signals. That is, in certain examples, the one or more user input signals are received that control the relationship between the estimated SNR and the noise reduction strength. The noise reduction strength parameter (i.e., the function relating SNR to noise reduction strength) is then adjusted based on one or more user input signals. As noted above, this optional user control of the noise reduction strength parameter/mixing ratio (shown in FIGs. 2 and 3 by dashed arrows 275 and 375) can adjust the relationship between the mixing ratio and the SNR in accordance with the recipient’s preferences. For example, the recipient could adjust the mixing ratio in different sound environments, with different sound types, etc. [0074] As should be appreciated, while particular uses of the technology have been illustrated and discussed above, the disclosed technology can be used with a variety of devices in accordance with many examples of the technology. The above discussion is not meant to suggest that the disclosed technology is only suitable for implementation within systems akin to that illustrated in the figures. In general, additional configurations can be used to practice the processes and systems herein and/or some aspects described can be excluded without departing from the processes and systems disclosed herein. [0075] This disclosure described some aspects of the present technology with reference to the accompanying drawings, in which only some of the possible aspects were shown. Other aspects can, however, be embodied in many different forms and should not be construed as Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 limited to the aspects set forth herein. Rather, these aspects were provided so that this disclosure was thorough and complete and fully conveyed the scope of the possible aspects to those skilled in the art. [0076] As should be appreciated, the various aspects (e.g., portions, components, etc.) described with respect to the figures herein are not intended to limit the systems and processes to the particular aspects described. Accordingly, additional configurations can be used to practice the methods and systems herein and/or some aspects described can be excluded without departing from the methods and systems disclosed herein. [0077] According to certain aspects, systems and non-transitory computer readable storage media are provided. The systems are configured with hardware configured to execute operations analogous to the methods of the present disclosure. The one or more non-transitory computer readable storage media comprise instructions that, when executed by one or more processors, cause the one or more processors to execute operations analogous to the methods of the present disclosure. [0078] Similarly, where steps of a process are disclosed, those steps are described for purposes of illustrating the present methods and systems and are not intended to limit the disclosure to a particular sequence of steps. For example, the steps can be performed in differing order, two or more steps can be performed concurrently, additional steps can be performed, and disclosed steps can be excluded without departing from the present disclosure. Further, the disclosed processes can be repeated. [0079] Although specific aspects were described herein, the scope of the technology is not limited to those specific aspects. One skilled in the art will recognize other aspects or improvements that are within the scope of the present technology. Therefore, the specific structure, acts, or media are disclosed only as illustrative aspects. The scope of the technology is defined by the following claims and any equivalents therein. [0080] It is also to be appreciated that the embodiments presented herein are not mutually exclusive and that the various embodiments may be combined with another in any of a number of different manners.

Claims

Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 CLAIMS What is claimed is: 1. A method, comprising: receiving at least one input signal at one or more input devices; processing the at least one input signal with a noise reduction process to generate a spectral gain for application to the at least one input signal; and determining a signal-to-noise ratio (SNR) of the at least one input signal based on the spectral gain. 2. The method of claim 1, wherein determining the SNR of the at least one input signal based on the spectral gain comprises: applying the spectral gain to the at least one input signal to generate at least one noise- reduced signal; and determining the SNR of the at least one input signal as a function of a ratio of the at least one input signal to the at least one noise-reduced signal. 3. The method of claim 2, further comprising: determining a noise reduction strength parameter as a function of the SNR; and mixing the at least one noise-reduced signal with the at least one input signal in accordance with the noise reduction strength parameter to generate an output signal. 4. The method of claim 3, further comprising: determining the noise reduction strength parameter as a function of the SNR and based on one or more user inputs. 5. The method of claim 1, wherein determining the SNR of the at least one input signal based on the spectral gain comprises: determining the SNR directly from the spectral gain. 6. The method of claim 5, further comprising: determining a noise reduction strength parameter as a function of the SNR; and generating an output signal from the spectral gain and the noise reduction strength parameter. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 7. The method of claim 6, further comprising: determining the noise reduction strength parameter as a function of the SNR and based on one or more user inputs. 8. The method of claim 1, 2, 3, 4, 5, 6, or 7, wherein determining the SNR of the at least one input signal based on the spectral gain comprises: generating an instantaneous SNR estimated based on the spectral gain. 9. The method of claim 1, 2, 3, 4, 5, 6, or 7, wherein determining the SNR of the at least one input signal based on the spectral gain comprises: determining a smoothed SNR estimated over a period of time. 10. The method of claim 1, 2, 3, 4, 5, 6, or 7, wherein determining the SNR of the at least one input signal based on the spectral gain comprises: determining a smoothed SNR estimated over a frequency range. 11. The method of claim 1, 2, 3, 4, 5, 6, or 7, wherein receiving the at least one input signal at one or more input devices comprises: receiving at least one sound signal at one or more sound inputs. 12. The method of claim 1, 2, 3, 4, 5, 6, or 7, wherein the at least one input signal is a time domain signal, and wherein the method comprises: converting the at least one input signal into a frequency domain representation using overlapping window frames prior to processing the at least one input signal with the DNN- based noise reduction process. 13. The method of claim 1, 2, 3, 4, 5, 6, or 7, wherein processing the at least one input signal with a noise reduction process comprises: processing the at least one input signal with a Deep Neural Network (DNN)-based noise reduction process. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 14. One or more non-transitory computer readable storage media comprising instructions that, when executed by one or more processors of a hearing device, cause the one or more processors to: calculate a noise-reduction gain for application to one or more sound signals received at sound inputs of the hearing device; determine a signal-to-noise ratio (SNR) of the one or more sound signals based on the noise-reduction gain; apply the noise-reduction gain to the one or more sound signals to generate one or more noise-reduced signals; and mix the one or more noise-reduced signals with the one or more sound signals in accordance with a mixing ratio, wherein the mixing ratio is a function of the SNR. 15. The one or more non-transitory computer readable storage media of claim 14, wherein the instructions that cause the one or more processors to calculate a noise-reduction gain comprise instructions that, when executed, cause the one or more processors to: process the one or more sound signals with a Deep Neural Network (DNN)-based noise reduction process to generate the noise-reduction gain. 16. The one or more non-transitory computer readable storage media of claim 14 or 15, wherein the instructions that cause the one or more processors to determine the SNR of the one or more sound signals based on the noise-reduction gain comprise instructions that, when executed, cause the one or more processors to: apply the noise-reduction gain to the one or more sound signals to generate the one or more noise-reduced signals; and determine the SNR of the one or more sound signals as a function of a ratio of the one or more sound signals to the one or more noise-reduced signals. 17. The one or more non-transitory computer readable storage media of claim 14 or 15, wherein the instructions that cause the one or more processors to determine the SNR of the one or more sound signals based on the noise-reduction gain comprise instructions that, when executed, cause the one or more processors to: generate an instantaneous SNR estimated based on the noise-reduction gain. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 18. The one or more non-transitory computer readable storage media of claim 14 or 15, wherein the instructions that cause the one or more processors to determine the SNR of the one or more sound signals based on the noise-reduction gain comprise instructions that, when executed, cause the one or more processors to: generate a smoothed SNR estimated over a period of time. 19. The one or more non-transitory computer readable storage media of claim 14 or 15, wherein the instructions that cause the one or more processors to determine the SNR of the one or more sound signals based on the noise-reduction gain comprise instructions that, when executed, cause the one or more processors to: generate a smoothed SNR estimated over a frequency range. 20. The one or more non-transitory computer readable storage media of claim 14 or 15, further comprising instructions that cause the one or more processors to: receive one or more user input signals adjusting a relationship between the mixing ratio and the SNR; and determine the mixing ratio based on the SNR and based on one or more user input signals. 21. One or more non-transitory computer readable storage media comprising instructions that, when executed by one or more processors of a hearing device, cause the one or more processors to: calculate a noise-reduction gain for application to one or more sound signals received at sound inputs of the hearing device; determine a signal-to-noise ratio (SNR) of the one or more sound signals based on the noise-reduction gain; determine a noise reduction strength parameter as a function of the SNR; and generate an output signal from the noise-reduction gain and the noise reduction strength parameter. 22. The one or more non-transitory computer readable storage media of claim 21, wherein the instructions that cause the one or more processors to calculate a noise-reduction gain comprise instructions that, when executed, cause the one or more processors to: Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 process the one or more sound signals with a Deep Neural Network (DNN)-based noise reduction process to generate the noise-reduction gain. 23. The one or more non-transitory computer readable storage media of claim 21 or 22, wherein the instructions that cause the one or more processors to determine the SNR of the one or more sound signals based on the noise-reduction gain comprise instructions that, when executed, cause the one or more processors to: generate an instantaneous SNR estimated based on the noise-reduction gain. 24. The one or more non-transitory computer readable storage media of claim 21, wherein the instructions that cause the one or more processors to determine the SNR of the one or more sound signals based on the noise-reduction gain comprise instructions that, when executed, cause the one or more processors to: generate a smoothed SNR estimated over a period of time. 25. The one or more non-transitory computer readable storage media of claim 21 or 22, wherein the instructions that cause the one or more processors to determine the SNR of the one or more sound signals based on the noise-reduction gain comprise instructions that, when executed, cause the one or more processors to: generate a smoothed SNR estimated over a frequency range. 26. The one or more non-transitory computer readable storage media of claim 21 or 22, further comprising instructions that cause the one or more processors to: receive one or more user input signals adjusting a relationship between the noise reduction strength parameter and the SNR; and determine the noise reduction strength parameter based on the SNR and based on one or more user input signals. 27. A device, comprising: one or more input devices configured to receive a noisy input signal, wherein the noisy input signal includes a target signal embedded in additive noise; a Deep Neural Network (DNN)-based noise reduction module configured to generate a DNN output signal from the noisy input signal; Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 a signal-to-noise ratio (SNR) estimation module configured to estimate an SNR of the noisy input signal based on the DNN output signal; a mapping module configured to generate a noise reduction strength function that relates the estimated SNR to a noise reduction strength; and an application module configured to use the noise reduction strength function to generate an output signal representing the target signal. 28. The device of claim 27, wherein the DNN output signal is applied to the noisy input signal to generate a noise-reduced signal, and wherein the SNR estimation module is configured: estimate the SNR of the noisy input signal based on a ratio of the noisy input signal to the noise-reduced signal. 29. The device of claim 27, wherein the SNR estimation module is configured to estimate the SNR directly from the DNN output signal. 30. The device of claim 27, 28, or 29, wherein the SNR estimation module is configured to generate an instantaneous SNR estimated based on the DNN output signal. 31. The device of claim 27, 28, or 29, wherein the SNR estimation module is configured to generate a smoothed SNR estimated over a period of time. 32. The device of claim 27, 28, or 29, wherein the SNR estimation module is configured to generate a smoothed SNR estimated over a frequency range. 33. The device of claim 27, 28, or 29, wherein the mapping module is configured to: receive one or more user input signals controlling the relationship between the estimated SNR and the noise reduction strength; and adjust the noise reduction strength function based on one or more user input signals. 34. The device of claim 27, 28, or 29, wherein the one or more input devices comprises sound input devices. Atty. Docket No.3065.0740i Client Ref. No. CID03712WOPC1 35. The device of 27, 28, or 29, wherein the DNN output signal is used to generate a noise-reduced signal, and wherein application module is configured to generate the output signal based on the noisy input signal, the noise-reduced signal, and the noise reduction strength function. 36. The device of 27, 28, or 29, wherein the application module is configured to generate the output signal based on the DNN output signal and the noise reduction strength function.