Electronic device and method

Abstract

An electronic device includes circuitry for noise-cancellation. The circuitry is configured to receive a reference ambient noise signal, predict a set of ambient noise signals for noise-cancellation based on the acquired reference ambient noise signal at a first point of time, select a predicted ambient noise signal from the set of ambient noise signals, based on the acquired reference ambient noise signal at a second point of time and perform noise-cancellation based on the predicted ambient noise signal.

Description

ELECTRONIC DEVICE AND METHODTECHNICAL FIELDThe present disclosure generally pertains to an electronic device and a method.TECHNICAL BACKGROUNDGenerally, noise cancelling techniques are known. For example, active noise cancellation (ANC) may be used for filtering out low background frequencies and is used in headphones. Such headphones may emit an anti-noise signal to compensate for the surrounding noise, such that the surrounding noise is minimized or even cancelled.Although there exist techniques for noise cancellation, it is generally desirable to improve on existing techniques.SUMMARYAccording to a first aspect the present disclosure provides an electronic device comprising circuitry for noise-cancellation configured to: receive a reference ambient noise signal; predict a set of ambient noise signals for noise-cancellation based on the acquired reference ambient noise signal at a first point of time; select a predicted ambient noise signal from the set of ambient noise signals, based on the acquired reference ambient noise signal at a second point of time; and perform noise-cancellation based on the predicted ambient noise signal.According to a second aspect the present disclosure provides a method for performing noisecancellation comprising: obtaining a reference ambient noise signal; predicting a set of ambient noise signals for noise-cancellation based on the acquired reference ambient noise signal at a first point of time; selecting a predicted ambient noise signal from the set of ambient noise signals, based on the acquired reference ambient noise signal at a second point of time; and performing noise-cancellation based on the predicted ambient noise signal.Further aspects are set forth in the dependent claims, the drawings and the following description.BRIEF DESCRIPTION OF THE DRAWINGSEmbodiments are explained by way of example with respect to the accompanying drawings, in which:Fig. 1 illustrates an embodiment of an electronic device for noise cancellation;Fig. 2 illustrates an embodiment of an electronic device for noise cancellation;Fig. 3 illustrates an embodiment of a set of predicted ambient noise signals compared to a reference ambient noise signal;Fig. 4 illustrates an embodiment of a set of predicted ambient noise signals compared to a reference ambient noise signal if the first timepoint of the reference signal used for prediction is the same as the second timepoint of the reference signal used for selection;Fig. 5 illustrates noise cancellation of an embodiment based on a predicted ambient noise signal;Fig. 6 illustrates noise cancellation of an embodiment based on a source separated predicted ambient noise signal;Fig. 7 illustrates a network including a server as well as electronic devices of an embodiment;Fig. 8 illustrates an embodiment of a method for noise cancellation; andFig. 9 illustrates an embodiment of an electronic device for noise cancellation.DETAILED DESCRIPTION OF EMBODIMENTSBefore a detailed description of the embodiments under reference of Fig. 1 is given, general explanations are made.As mentioned in the outset, some noise cancelling techniques are generally known. However, general noise cancellation may not always be effective. It has been recognized that state of the art active noise cancellation algorithms only consider the instantaneous noise and compute the anti-noise signal in real time to suppress it before it reaches the ear. This feed-forward approach may sometimes be extended with (error-)correction microphone signals, which may control malicious behavior. However, it has been recognized that only taking into consideration the instantaneous noise may be problematic. Instead, some of the present embodiments relate to predicting future noise and taking future noise into consideration for noise cancellation. Thus, some embodiments may relate to computing an anti-noise signal for future events in order to reduce the noise amplitude further.Hence, some embodiments pertain to an electronic device comprising circuitry for noisecancellation configured to: receive a reference ambient noise signal; predict a set of ambient noise signals for noise-cancellation based on the acquired reference ambient noise signal at a first point of time; select a predicted ambient noise signal from the set of ambient noise signals, based on the acquired reference ambient noise signal at a second point of time; and perform noise-cancellation based on the predicted ambient noise signal.The electronic device may be headphones, earphones (e.g., in-ear headphones), a smartphone, a laptop computer, a tablet computer, a personal computer, a wearable electronic device (e.g., electronic glasses, etc.), a vehicle (e.g., a car), or the like. The electronic device may include one or more microphones. For example, one or more microphones for every ear, such as one or more microphones, e.g., reference microphones, for picking up ambient noise signals, i.e., audio signals from the environment. Also, one or more error microphones for measuring the effectiveness of the noise cancellation may be included in the electronic device, such as microphones facing towards the user’s ear for measuring the remaining sound reaching the ear after noise cancellation. An error microphone may direct towards a user’s ear canal and measure the resulting noise level closest to the eardrum, for example, the error microphone may measure the sound signal between the electronic device and the user’s ear. One error microphone for each ear may be included in the electronic device. The error microphone may measure the remaining sound between the outside surface of the electronic device and the user’s eardrum, e.g., in regard to headphones or the like this may refer to the area between the headphone membrane and the user’s eardrum. The electronic device may include one or more microphones for switch-control, for example, located closely in front of the loudspeaker emitting the anti-noise. The error microphone may be used for switch-control. The electronic device may include one or more loudspeakers. For example, one or more loudspeakers for every ear. One or more loudspeakers may be used for emitting the anti-noise signal. For example, one loudspeaker for each ear may emit an anti-noise signal. One or more loudspeakers, which may be the same as the one emitting the anti-noise signal, may also be used for playing other audio signals, such as music or phone calls or the like.The circuitry may include one or more entities capable of processing audio signals, such as a CPU (central processing unit), GPU (graphics processing unit), FPGA (field programmable gate array), an application-specific integrated circuit (ASIC), and / or any suitable kind of programmable microprocessor or integrated circuit and / or any other kind of processor or the like.A functionality of the circuitry may be specified at least in part by a hardware configuration of the circuity and / or at least in part by software stored on or provided to the circuitry, wherein the software may include instructions executed by the circuitry.The circuitry may also include a storage, a memory (RAM, ROM or the like) that, for example, stores (e.g., temporary) data or signals that are generated for processing audio signals. The memory may include a flipflop, a latch, a static random-access memory (SRAM), an embedded dynamic random-access memory (eDRAM) or the like. The circuitry may further include acommunication unit for receiving signals (e.g., the reference ambient noise signal), and / or for outputting signals (e.g., the anti-noise signal). The communication unit may include a processor pin, a peripheral component interconnect (PCI) interface, a universal serial bus (USB) interface, or the like.Further, the circuitry may include input means (mouse, keyboard, camera, etc.) and / or output means (display (e.g., liquid crystal, (organic) light emitting diode, etc.), loudspeakers, etc., a (wireless) interface, etc., for example, an interface as it is generally known for electronic devices (computers, smartphones, etc.). Moreover, it may include sensors for sensing audio signals, etc.The reference ambient noise signal and the predicted ambient noise signal may refer to an audio signal. An audio signal may correspond to a wired or wireless signal which can be translated or converted into sound and, in some embodiments, which may be based on sound. However, the sound may be real or completely artificial or simulated.The audio signal may be both wired and wireless. For example, the reference ambient noise signal may be obtained by the electronic device, for example, by the circuitry, and may be processed as an electric signal, for example by the circuitry, but before or after the audio signal is processed by the circuitry, the audio signal may be transmitted wirelessly, e.g., via an air interface. Similarly, an anti-noise signal may be transmitted wirelessly. Also, a microphone of the electronic device may capture the reference ambient noise signal. Hence, the reference ambient noise signal may be based on real sound, for example captured by the microphone of the electronic device, or may be transmitted based on a radio signal, or the like, for example, to a communication unit of the electronic device.Hence, obtaining or receiving an audio signal, e.g., the reference ambient noise signal, may refer to passively receiving or actively acquiring the audio signal.The reference ambient noise signal may refer to the ambient sound in the environment. Ambient sound may be any sound of the environment, for example, the sounds of objects, animals and persons, it may also refer to speech of persons etc.Predicting a set of ambient noise signals for noise cancellation based on the acquired reference ambient noise signal at a first time point may refer to predicting one or more ambient noise signals. That is, the set of ambient noise signals that is predicted may include one or more signals.Furthermore, the acquired reference ambient noise signal may be continuously acquired over time. The prediction of the set of ambient noise signals may be based on only part of the wholesignal, for example, only the part of the signal acquired until the first time point may be the basis for the prediction. That is, after the first time point the reference ambient noise signal may continue to be acquired. Of course, another cycle of prediction of the set of ambient noise signals may start with a new first timepoint.Predicting the set of ambient noise signals may refer to predicting the continuation of the reference ambient noise signal beyond the first timepoint. The electronic device may use the current reference ambient noise signal, i.e., the reference ambient noise signal at the first timepoint, which may be the current input of an external microphone of the electronic device (headphones, phone, vehicle etc.), and based on a short sequence of the noise captured, e.g., based on the reference ambient noise signal at the first point of time, the electronic device may predict the continuation of the audio sequence (sound wave) in the near future, i.e., the continuation of the ambient noise signal beyond the first timepoint. The near future may refer to 10ms, 50ms, 100ms and / or 1000ms. That is the reference ambient noise signal may be predicted up to 10ms, up to 50ms, up to 100ms and / or up to 1000ms beyond the first timepoint.Selecting the predicted ambient noise signal from the set of ambient noise signals may refer to selecting one signal from the set of signals. For example, if the set of signals (predicted set of ambient noise signals) includes more than one signal, selecting may refer to choosing one signal of the multiple signals in the set.Selecting the predicted ambient noise signal may be based on comparing the set of ambient noise signals to the acquired ambient noise signal at a second point of time. That is, each signal of the set of signals may be compared to the reference ambient noise signal. Based on the result of the comparison the predicted ambient noise signal may be selected from the set as a basis for the noise cancellation. For example, the signal of the set with the least difference to the reference signal (reference ambient noise signal) may be selected as the predicted ambient noise signal on which the noise cancellation is based.Thus, in the timesteps after the prediction, i.e., also the time after the first timepoint of the acquired reference ambient noise signal, the electronic device may check the actual ambient noise signal vs the predicted ambient noise signals of the set to determine the closest fit to the real ambient noise and utilize it, i.e., the one predicted ambient noise signal of the set with the closest fit, for noise cancellation, for example active noise cancellation.Furthermore, selecting the predicted ambient noise signal may be based on selecting from the compared set of ambient noise signals the signal if the difference to the acquired reference ambient noise signal at a second point of time is below a threshold. For example, a threshold ofdifference may be predefined. Then, when comparing each signal in the set of signals to the reference signal the difference between each predicted signal of the set to the reference signal may be determined. If the difference falls below a threshold the signal may be selected as the predicted ambient noise signal as a basis for the noise cancellation. If multiple signals fall below threshold than the one signal with the smallest difference may be selected.Selecting the predicted ambient noise signal may be based on comparing a part of the set of ambient noise signals corresponding to the acquired reference ambient noise signal at a second point of time. That is, the reference signal may include a signal on which basis the prediction of the set of ambient noise signals is made, but it may include also other parts. For example, if the reference signal is a continuously obtained signal, the prediction may be based on the reference signal acquired until a first time point. However, the reference signal may continue to be acquired after the first time point.The first point of time may occur at the same time or earlier than the second point of time.If the first point of time and the second point of time are the same, the selection of the predicted ambient noise signal from the set of signals, e.g., including the comparison as described above, may be based on that part of the reference ambient noise signal on which the prediction of the set of signals is based, i.e., the reference signal acquired until the first timepoint.If the first time point occurs earlier than the second time point, the selection as described above, which may include a comparison as described above, may be based on a different part of the reference signal than the one used for the prediction of the set of ambient noise signals. That is, the selection or comparison may be based on a later part of the reference signal than the one used for the prediction.The prediction of the set of ambient noise signals may include a prediction in time, that is, a prediction how the reference ambient noise signal would continue beyond the first timepoint, i.e., beyond the time point of the reference signal used for the prediction, that is, later than the first timepoint. The set of signals may include a signal beyond the first timepoint and may also include as signal beyond the second timepoint.Performing the noise-cancellation may be based on generating and emitting an anti-noise signal based on the predicted ambient noise signal. The anti-noise signal may be 180°-phase inverted to the predicted ambient noise signal. Thus, the anti-noise signal may reduce or cancel the ambient noise arriving at the ear of a user. That is, if the anti -noise signal is 180°-phase inverted to the predicted ambient noise signal it may also be essentially 180°-phase inverted to the ambientnoise (soundwave) in the environment on which basis the predicted ambient noise signal was generated.The anti -noise signal may be emitted to the user’s ear from a loudspeaker of the electronic device. The above described process may be conducted for each ear and an anti-noise signal may be emitted from a loudspeaker close to each ear of the user. The emitted anti-noise signal may be emitted at a time so that it is essentially 180°-phase inverted to the ambient noise signal arriving at the ear. Therefore, when the ambient noise and the anti-noise signal, which is essentially the 180°-phase inverted ambient noise, arrive at the ear of the user, the two signals cancel each other out and the user does not hear the ambient noise.The circuitry may be further configured to error measure an audio signal after the anti-noise signal was emitted for the noise-cancellation. That is, between the loudspeaker emitting the antinoise signal and the ear of the user an error microphone may be located to measure the remaining signal. The error microphone may be facing in the direction of the user’s ear. If the anti -noise signal completely cancels or highly reduces the ambient noise travelling to the ear, the error microphone may not pick up any audio signal or may only pick up a very low remaining signal. For example, the remaining signal may be below a predetermined threshold to indicate a sufficient noise cancellation or above the predetermined threshold to indicate insufficient noise cancellation. Thus, the error measured audio signal may refer to the remaining signal.The circuitry may be further configured to switch from noise-cancellation based on the predicted ambient noise signal to noise-cancellation based on the reference ambient noise signal based on the error measurement. The switching may occur if the error measurement is above a predetermined threshold. Noise-cancellation based on the reference ambient noise signal without the predicted ambient noise signal may also be referred to as reactive noise-cancellation. Thus, the circuitry may be configured to switch from noise-cancellation based on the predicted ambient noise signal to reactive noise-cancellation. The switch may be based on the error measurement. Similarly, a switch back, i.e., from reactive noise cancellation to noise-cancellation with prediction, may occur, which may also be based on the error measurement.That is, if the error microphone picks up an audio signal it may mean that the anti-noise signal may not completely cancel the ambient noise arriving at the user’s ear. If the remaining signal picked up by the error microphone is above a predetermined threshold it may indicate insufficient noise cancellation. Thus, the noise cancellation based on the predicted ambient noise signal may not be sufficient. In that case, the noise cancellation may be switched from the noise cancellation based on prediction to traditional noise cancellation without prediction. That is, intraditional noise cancellation the anti-noise signal is not based on prediction of the audio signal, but instead it is based on a reference signal without prediction.In other words, if the prediction does not correspond to the sound arriving at a user’s ear the noise cancellation may be switched from noise cancellation with prediction to noise cancellation without prediction. In the same vein, a switch back from noise cancellation without prediction to a noise cancellation with prediction may be performed based on the error measurement.The circuitry may be further configured to perform error correction based on the error measurement. Error correction may refer to adjustment in phase of the anti-noise signal to match the ambient sound arriving at the user’s ear. That is the anti -noise signal may be an essentially inverted ambient noise, but it may include an offset from the ideal 180°-phase inversion relative to the ambient noise arriving at the user’s ear. The offset may be recognized by the error measurement and corrected.In some embodiments, the circuitry may be further configured to perform source separation of the reference ambient noise signal, wherein predicting the set of ambient noise signals may be based on a source separated ambient noise signal.The circuitry may be further configured to merge at least two predicted ambient noise signals of different sources, which are based on source separated sets of ambient noise signals, wherein the noise-cancellation may be based on the merged predicted ambient noise signal.Given that typically there are multiple noises in the typical environment (e.g., urban, transportation, rural, etc.), sound separation of these noises may be utilized. Captured ambient noises, i.e., the different sources included in the acquired reference ambient noise signal, could be separated and classified (e.g., car engine, wind, talking / shouting, birds chirping etc.). The prediction of the set of ambient noise signals may then be performed based on each source. The predicted sounds, i.e., the predicted ambient noise signals of different sources, may be merged together for conducting noise cancellation based on the merged signal. That is, the anti-noise signal may be based on the merged signal.In some embodiments, the circuitry may be further configured to obtain a location-based noise profile, wherein performing the prediction may be based on the location-based noise profile. Obtaining the location-based noise profile may be based on crowdsourcing of multiple users.Thus, a context of the environment may be estimated by the circuitry. The context of the environment may be a location-based context (location-based profile). Depending on the estimated context, the predicted sounds, i.e., the predicted ambient noise signals of differentsources, may be combined back together (merged) to provide the full predicted ambient noise for the active noise cancellation input. The merging may for example occur before or after selection. That is, for example, the set of signals may be merged before selection or the selected predicted ambient noise signals may be merged after selection.For example, the user may live close to roadworks. In that context of the environment, the jack hammer may always make the same wideband noise which is an easily identifiable pattern of sound. The predictor predicting the set of ambient noise signals (the predictor may for example be based on generative artificial intelligence, see below) may easily generate the portion of the sound that is yet to come just from the beginning sequence, i.e., may easily predict the set of ambient noise signals related to the jack hammer source, wherein the set of ambient noise signals include the predicted future sound signal of the jack hammer source, based on the earlier reference signal of the jack hammer source. As the pattern of the jack hammer is continuously the same wideband noise over a period of time, the prediction is easy (in particular prediction via Al is easy, see below regarding explanation of generative Al).As the jack hammer is usually included in the roadworks, the user living close to the roadworks may always need a sound cancellation based on the jack hammer source sound pattern if they are at home at a certain period of time in the day. Thus, the predicted sound signal of the jack hammer source may always be included in the merged signal for noise cancellation for the user located at home at a certain time of day. Therefore, based on the context, e.g., location-based, but also time-based, a profile of the environment may be estimated which may influence the merging of the predicted ambient noise signals on which basis the noise cancellation is performed.The circuitry may be further configured to: receive a second (or more) reference noise signal associated with the reference ambient noise signal; and predict a set of second noise signals based on the acquired second reference noise signal; wherein performing noise-cancellation may include selecting a predicted second noise signal of the set of second noise signals based on the predicted ambient noise signal and performing noise-cancellation based on the predicted second noise signal.Thus, instead of one reference ambient noise signal also multiple reference ambient noise signals may be received by the circuitry as a basis for prediction. The second reference noise signal may refer to one of the multiple reference ambient noise signals. However not only a second reference noise signal also a third, fourth or nthreference noise signal may be received, all of which may be one of the multiple reference ambient noise signals. Each of the multiple reference ambient noise signals may be the basis for predicting its own set of ambient noise signals.Alternatively, one set of ambient noise signals may be based on multiple reference ambient noise signals. If a set of ambient noise signals is generated for each reference ambient noise signal a predicted ambient noise signal may be selected for each set of ambient noise signals and noise cancellation may be based on each predicted ambient noise signal. In this way, if multiple reference ambient noise signals are received also multiple predicted ambient noise signals may be generated. The processing of each of the multiple reference ambient noise signals, that is the prediction, selection and noise cancellation based on the multiple ambient noise signals may correspond to the processing, e.g., prediction, selection, noise cancellation, as described above and in the following for one reference ambient noise signal. Accordingly, any feature described in this specification with respect to one reference ambient noise signal may apply also to the multiple reference ambient noise signals, e.g., each of the multiple reference ambient noise signals, such as the second reference noise signal.The second reference noise signal may be an audio signal as explained above. The second reference noise signal may be captured by a second microphone, for example the error microphone. The second reference noise signal may be based on sound between an outer surface of the electronic device, e.g., the membrane of headphones, and the ear of the user. Each of the multiple reference ambient noise signals may be captured by a different microphone.The reference ambient noise signal and the second reference noise signal may be captured at the same time and may correspond to noise picked up at different locations in the environment. Thus, the reference ambient noise signal and the second reference noise signal, i.e., multiple reference ambient noise signals, may be based on sound of the same source captured at different locations. In other words, the second reference noise signal used for the prediction of the set of second noise signals may be the second reference noise signal at the first point of time, which may be the same first point of time of the reference ambient noise signal used for the prediction of the set of ambient noise signals. This may also apply to any of the multiple reference ambient noise signals. That is, multiple reference ambient noise signals may be based on sound of the same source captured at different locations in the same environment. As multiple reference ambient noise signals, e.g., including the second reference noise signal, may be captured from different points in the environment producing the ambient noise, e.g., captured from multiple different microphones, a library of reference ambient noise signals and in turn a library of their predictions may be generated. Therefore, similar reference ambient noise signals may be clustered based on location and their corresponding predictions may be used for noise cancellation. Clustering may for example be based on locations within a predetermined distance threshold from each other, such that the corresponding ambient noise captured in the respectivelocations is based on the same environment. Similarity of reference ambient noise signals may refer to corresponding reference ambient noise signals based on the same environment as the source of the ambient noise. In this way, the prediction of the set of ambient noise signals and / or the selection of the predicted ambient noise signal from the set for performing noise cancellation may not need to be performed each time noise cancellation is performed, but still the prediction can be improved over time by accumulating an increasing number of reference signals and their corresponding predictions for the library. That is, noise-cancellation may be based on the library of reference ambient noise signals and / or the library of predicted ambient noise signals.Thus, as the prediction is not needed to be calculated each time, processing power required for the noise cancellation system can be saved.Furthermore, the prediction performance, i.e., the prediction accuracy, increases with every additional reference ambient noise signal. For example, if the predicted ambient noise signal is generated based on generative artificial intelligence (Al), as described in more detail below, then the predicted ambient noise signal will be improved for noise cancellation if it is based on more information. An improved noise cancellation may refer to an increase in noise attenuation, i.e., a minimizing of the remaining noise after noise cancellation. In other words, the more information the predictor receives for the prediction the more accurate the predicted ambient noise signal can be. Receiving two or more reference ambient noise signals (of different locations, but based on the same sound), i.e., basing the prediction on multiple reference ambient noise signals, increases the information, such as the information about the sound source, as compared to receiving only a single reference ambient noise signal.In the following, prediction and noise cancellation based on the reference ambient noise signal in relation to the second reference noise signal are described. It is noted that any feature described for the reference ambient noise signal and the second reference noise signal may apply also in relation of any of the multiple reference ambient noise signals to each other.The prediction of the set of second noise signals may correspond to the prediction of the set of ambient noise signals based on the reference ambient noise signal as described above, and the prediction of the set of second noise signals may accordingly exhibit any feature described above (and any feature described below, e.g., concerning Al, concerning Figs. 1 to 9) with respect to the prediction based on the reference ambient noise signal, wherein instead of the reference ambient noise signal the second reference noise signal is used.For example, the prediction may include a prediction of the future of the second reference noise signal, e.g., a prediction beyond a first timepoint of the second reference noise signal used for prediction of the set of second noise signals may be performed.Also, for example, the anti-noise signal may be based on the predicted second noise signal. The predicted second noise signal may be based on the predicted ambient noise signal. That is, the set of second noise signals may include signals that correspond to the signals of the set of ambient noise signals, because the two reference signals on which the two sets are based may be associated with each other. The two reference signals may not only be associated with each other based on time, but also based on the noise source they are based on. That is, the noise source may produce noise that may be picked up by two microphones, the reference microphone for measuring the reference ambient noise signal and a second microphone, such as the error microphone for example.Thus, predictive sound for active noise cancellation may be generated for two locations of the environment.Concerning the reference ambient noise signal, sound may be based on the ambient environment, which may be captured by a reference microphone, for example, an outward facing microphone in headphones. The sound based on the ambient environment may for example be traffic noise, construction noise, etc.On the other hand, concerning the second reference noise signal (or also any other of the multiple reference ambient noise signals), sound may be based on the same ambient environment but captured by a different microphone. Additionally, the second reference noise signal may include sound based inside the electronic device, i.e., an inside noise signal may be included in the second reference noise signal. The sound based inside the electronic device may for example be based on the user manipulating the headphone resulting in noise, such as squeaking or rubbing noises or the like. Inside may refer to the area between the electronic device’ outer surface, in regard to headphones this may be the headphone membrane, and the user’s ear drum. The sound based inside the electronic device may be captured by an error microphone.The inside signal may be separated from the second reference signal based on sound separation. Sound separation may be performed as described above regarding sound separation of reference ambient noise signal based on multiple noises in the environment. For example, before the prediction the second reference signal may be decomposed, such that the inside signal is discarded and only the signal based on the ambient noise of the environment remains. Sound separation of the second reference noise signal may be based on the reference ambient noisesignal, which can function as a reference for source separating the inside signal from the rest of the second reference signal, i.e., form the signal based on ambient noise. In this way, the reference ambient noise signal and the second reference noise signal can be aligned so that both include only sound based on the same sound source, i.e., the ambient noise signal.As explained above an environment may include multiple noise sources, e.g., traffic noise, construction noise and also noise based inside the electronic device etc., thus, multiple reference ambient noise signals may correspond to multiple noise sources in the environment, but there may also be an overlap, wherein the multiple ambient reference signal are based on noise of the same source in the environment.The multiple reference ambient noise signals, e.g., the reference ambient noise signal and the second reference noise signal, may be used for generating the location-based profile of the environment as described above.Respective sets of predicted signals may be generated, i.e., the set of ambient noise signals and the set of second noise signals may be predicted respectively, and each set may include a few options, i.e., a few predicted signals.At a future timestep (e.g., second point of time) the actual ambient sound (i.e., reference ambient noise signal) vs the predicted sound (i.e., predicted set of ambient noise signals) may be compared to check if the prediction matches the reality, e.g., if the difference between at least one of the signals of the predicted set of ambient noise signals and the reference ambient noise signal is below a predetermined threshold. If that is the case, e.g., if at least one of the predicted set of ambient noise signals, which may for example be based on the reference ambient noise signal at a first point of time, matches the actual ambient sound at a future timestep (e.g., the reference ambient noise signal at a second point of time), the at least one matching signal of the predicted set of ambient noise signals may be used for noise cancellation, e.g., active noise cancellation. If not, the respective prediction of the same sound in the set of second noise signals, i.e., a corresponding predicted second noise signal of the set of second noise signals, may be used for noise cancellation, e.g., active noise cancellation. In other words, the two sets, the predicted set of second noise signals and the predicted set of ambient noise signals, may include corresponding predicted signals. However, if the set of second noise signals does not include a signal corresponding to the predicted ambient noise signal selected from the first set of ambient noise signals, then a switch to “reactive” noise-cancellation may be performed. This may occur, for example, if the reference ambient noise signal and the second reference noise signal do not correspond, for example because of the inside signal interfering with the second reference noisesignal. If the actual ambient sound (e.g., the reference ambient noise signal at a second point of time, the second reference noise signal (at a second point of time) etc.) and the predicted sound (e.g., the predicted set of ambient noise signals, predicted set of second noise signals etc.) do not match, e.g., if none of the signals of the predicted set of signals (e.g., set of ambient noise signals, set of second noise signals etc.) fall below a predetermined threshold of difference to the reference signal (e.g., reference ambient noise signal, second reference signal etc.) the system may switch to the traditional active noise cancellation, i.e., reactive active noise cancellation. Thus, if both predictions (in case of a predicted set of ambient noise signals and a predicted set of second noise signals) or all predictions (in case of multiple predicted sets of ambient noise signals) fail, the switch to reactive noise cancellation may occur.The multiple reference ambient noise signals, such as second reference noise signal, may be captured from microphones of different devices, which may be devices of different users. Thus, as explained above the multiple reference ambient noise signals may be used for crowdsourcing the location-based profile of the environment. That is, over time multiple reference ambient noise signals not only from the same user, but from multiple users, mas be used for generating the library of reference ambient noise signals and in turn the library of their predictions. Also, as describe above for multiple reference ambient noise signals, beside the reference ambient noise signal and the second reference noise signal, also further noise signals may be used as input for one or more predictions, which may be additional predictions. For example, signals from one or more additional microphones, which may pick up the sounds of the environment may be obtained. The prediction may thus not only be based on generating one or two sets of signals, but depending on the input even more than two sets of predicted signals may be generated.Also, in case of multiple reference ambient noise signals, e.g., captured by different microphones at different locations of the same environment, the prediction of a set of reference ambient noise signals for one location may be based on all of the multiple reference ambient noise signals. That is, the input of all microphones may inform the prediction of one set of reference ambient noise signals for a given location. In this way, for example, if there are multiple microphones, e.g., e.g., mic l, mic_2, . . ., mie n, each located at a different location in the same environment and each picking up the ambient noise from the same source as different corresponding reference ambient noise signals, e.g., xmic_i, xmic_2, . . . , xmic_n , and if the function of the predictor, e.g., the generative Al, is f, the sound continuation prediction at the location of mic l may be based on f(xmic 1, Xmic 2, . . . , xmic n), i.e., it may be based on the multiple reference ambient noise signals of the multiple microphones located at the multiple locations. Essentially, every microphone signal can be used to inform the prediction at each microphone location.Thus, some embodiments may pertain to one or more simultaneous predictions of the anti-noise, e.g., the prediction and selection of the closest fitting predicted noise signal as compared to a reference signal.Performing the prediction, i.e., the prediction of the set of ambient noise signals and / or the set of second noise signals, may be based on generative artificial intelligence (Al). It has been recognized that generative Al has been revolutionizing image and language generation. For example, large-language models (LLMs) and diffusion models which are based on generative Al may generate unseen images, expand viewpoints (filling unseen parts of the image) and even generate short video from a single image by utilizing context learned inside the model.Thus, in some embodiments the capabilities of generative Al may be utilized to generate audio sequences of various noises present in the human environment (sound predictions) and utilize these sound predictions for active noise cancellation (ANC), e.g., for active noise cancellation algorithms.Given that the prediction by generative Al may not be deterministic, a few options, i.e., a few variants of the predicted signal, may be predicted. That is, the set of ambient noise signals may include a few variants of predicted ambient noise signals. The same applies to any other set of predicted signals, for example the set of second noise signals.The predicted signals, e.g., the set of ambient noise signals, predicted by the generative Al, that is, for example, the variants of the predicted ambient noise signal, may be computed and stored on the electronic device. For example, the circuitry may include an Al processor to compute the set of signals via generative Al, or the CPU or any other processor of the circuitry may run the generative Al to predict the set of ambient noise signals. The set of ambient noise signals may then be stored in the storage or memory, for example, RAM, ROM etc.The generative Al may include or be based on an artificial neural network, e.g., a deep neural network, a convolutional neural network, a Transformer, a variational autoencoder (VAE), a generative adversarial network (GAN), or the like. For example, the generative Al may use or may be configured similar to DALL-E, Midjourney, Stable Diffusion, Deep Dream Generator, StyleGAN, BigGAN, or the like.Furthermore, generative Al may include a diffusion model with attention heads and adversarial loss.The generative Al may be trained in advance. For example, audio signals and / or anti-noise signals may be used for training the generative Al.Also, the generative Al may leverage multimodal input, e.g., the reference ambient noise signal and the second reference noise signal and / or even other signals may be used for prediction. The multimodal input may be transmitted from multiple microphones, for example, multiple microphones included in the electronic device. Also, an additional microphone, e.g., the error microphone, may be used as a switch-control, e.g., as described above, to enhance the noise cancellation.Concerning the location-based noise cancellation described above, the generative Al may be trained to adapt to the context, for example the location or time of day, of the user’s environment for cancelling the context specific, e.g., position and / or time specific, noises. This precise understanding of the semantics would allow a better time-aligned noise cancellation, e.g., with a smaller offset from the ideal 180°-phase inverted anti-noise signal to the ambient noise, i.e., the noise arriving at a user’s ear. Therefore, potentially a wider frequency range may be covered. In other words, the more is known about the environment and the available sound sources of the environment, such as the typically available sound sources, the better the prediction of (multiple) anti-noise signals will be. The anti-noise signals can, thusly, have a broader frequency range and therefore cancelling at perfect anti-phase alignment can be achieved.The classification of location-based noises, i.e., the location-based sound sources, may be crowdsourced from multiple users using an enabled electronic device of the present embodiments. The multiple users may share, for example, via a database, the location-based sound sources among everyone, for example, every user, approaching the vicinity of the location, wherein the vicinity may be based on a predetermined threshold.The training of the generative Al may, for example, be based on locations and / or times as well as their corresponding audio signals and audio sources.Thus, some embodiments may pertain to classifying noise sources to train the prediction model, e.g., the generative Al for predicting the set(s) of predicted signals. The knowledge about the environment, may be an additional feature of the prediction model.Some embodiments pertain to a method for performing noise-cancellation comprising: obtaining a reference ambient noise signal; predicting a set of ambient noise signals for noise-cancellation based on the acquired reference ambient noise signal at a first point of time; selecting a predicted ambient noise signal from the set of ambient noise signals, based on the acquired reference ambient noise signal at a second point of time; and performing noise-cancellation based on the predicted ambient noise signal.The method may correspond to the electronic device and circuitry described above, and the method may accordingly exhibit any feature described above with respect to the electronic device and / or circuitry and / or any suitable feature described below with respect to any one of the figures. The method may be performed by the circuitry and / or by the electronic device described above.For example, concerning the method, selecting the predicted ambient noise signal may be based on comparing the set of ambient noise signals to the acquired ambient noise signal at a second point of time. Selecting the predicted ambient noise signal may be based on selecting from the compared set of ambient noise signals the signal if the difference to the acquired reference ambient noise signal at a second point of time is below a threshold. Also, selecting the predicted ambient noise signal may be based on comparing a part of the set of ambient noise signals corresponding to the acquired reference ambient noise signal at a second point of time, wherein the first point of time may occur at the same time or earlier than the second point of time.The methods as described herein are also implemented in some embodiments as a computer program causing a computer and / or a processor to perform the method, when being carried out on the computer and / or processor. In some embodiments, also a non-transitory computer- readable recording medium is provided that stores therein a computer program product, which, when executed by a processor, such as the processor described above, causes the methods described herein to be performed.Returning to Fig. 1, an embodiment of an electronic device for noise cancellation is illustrated. The electronic device 1 represents headphones including headphone membrane la facing an ear (not visible) of a user, reference microphone 3 for capturing ambient noise signal 2, loudspeaker 4 for emitting the anti-noise signal and error microphone 5 for measuring the error signal and adapting the noise cancellation by either adjusting an offset of the anti-noise signal to the ideal 180°-inverted ambient noise signal 2 and / or for switching from the noise cancellation based on prediction to noise cancellation without prediction. The electronic device 1 predicts the set of ambient noise signals based on ambient noise signal 2 captured by microphone 3. The set of ambient noise signals is compared to the captured ambient noise signal 2 and the predicted ambient noise signal of the set with the closest fit to the ambient noise signal 2 is selected for noise cancellation, on which basis the anti-noise signal is generated and emitted from loudspeaker 2. Error microphone 5, located between loudspeaker 4 and an ear of the user (not visible), measures the remaining signal and performs error correction if necessary for removing an offset as describe above. Additionally, if the remaining signal, either before and / or after errorcorrection, is above a predetermined threshold a switch may occur in a way that the generation of the anti-noise signal is no longer based on the predicted set of signals, i.e., the predicted ambient noise signal, but instead on the basis of the captured ambient noise signal without prediction. Another switch, i.e., a switch back, may occur later on based on the error measurement of error microphone 5. The prediction, selection and switch is also explained in more detail in Fig. 2.Fig. 2 illustrates an embodiment of an electronic device for noise cancellation. The sound waves of an ambient noise signal 2 generated by noise source 21 arrive at microphone 3 of the electronic device 1, which may be the headphones of Fig. 1, and are captured as ambient noise signal 2 at the first timepoint tl (see Figs. 3 and 4) by microphone 3. The ambient noise signal 2 is transmitted to the determination section 10 of the electronic device 1, wherein further processing of the ambient noise signal 2 captured at the first timepoint occurs. In the meantime, the soundwaves of the ambient noise signal 2, after arrival at the microphone 2 do not stop and travel further in the direction of the ear 6 of a user as illustrated by arrow 16. Also, microphone 3 does not stop capturing ambient noise signal 2 even beyond the first timepoint (tl, Figs. 3 and 4). While the noise soundwaves travel forward, as indicated by arrow 16, the generative artificial intelligence model 15 included in the determination section 10 generates a set of predicted ambient noise signals based on the captured ambient noise signal 2 at the first time point. The predicted set of ambient noise signals include multiple signals each of which predict how the ambient noise signal 2 will develop beyond the first timepoint, for example 10ms, or 50ms, or 100ms, or 1000ms beyond the first timepoint. That is, each of the signals of the set may predict to the same timepoint in the future (10ms, or 50ms, or 100ms, or 1000ms beyond the first timepoint). The prediction may be even beyond a second timepoint as explained below. While the set of predicted ambient noise signals is generated, reference microphone 3 continues to capture the reference ambient noise signal 2 at a second timepoint tl+n (see Fig. 3) after the first timepoint tl and transmits the reference ambient noise signal 2 at a second time point tl+n after the first timepoint tl to the determination section 10. Each of the signals of the set of signals is compared to the reference ambient noise signal 2. The comparison may be based on the part of the signals starting from the start of the signals to the first timepoint tl and / or from the first timepoint tl to the second timepoint tl+n. That is, even though the set of predicted ambient noise signals may continue beyond the second timepoint tl+n, the reference signal may not yet have been acquired beyond the second timepoint tl+n. Therefore, the comparison may be conducted for the section of signals from timepoints tl to tl+n, as in this way, the predicted signal from tl to tl+n may be compared to the actual signal captured by reference microphone 3 from tl to tl+n. Based on the comparison, the signal of the set with the closest match to the referenceambient noise signal 2 is selected as the predicted ambient noise signal on which basis the antinoise signal is generated, which is transmitted to loudspeaker 4 which emits the anti-noise signal 14 as soundwaves. In the meanwhile, the noise soundwaves corresponding to the reference ambient noise signal 2 have travelled to location 18 corresponding to the start, i.e., emission point, of the emitted anti-noise signal 14. Thus, the anti-noise signal 14 and the ambient noise signal 2 travel in the direction of the ear 6, as indicated by arrows 22 and 16, and cancel each other out on the way, as the anti -noise signal may be an essentially 180°-phase inverted signal of the noise soundwaves, possibly with an offset. The error microphone 5 included in electronic device 1 and located between loudspeaker 4 and ear 6 measures how well the noise cancellation works and whether there remains a signal after noise cancellation. This measurement of the error microphone 5, the error measurement, is transmitted to the determination section 10 and influences the generation of the anti-noise signal 14. That is, error correction of the anti-noise signal 14 may occur for the anti-noise signal in a way that its noise cancellation capabilities are improved, i.e., in a way that a phase offset to the ideal of a 180°-phase inversion of the noise sound waves at point 15, may be compensated, e.g., minimized. Also, if error correction is not enough, that is, if the remaining signal measured by error microphone 5 and / or corrected via error correction is above a predetermined threshold, a switch to noise cancelling without prediction may occur. In that case the anti-noise signal 14 is based on the currently measured reference ambient noise signal 2, but not on a predicted future signal. The switch may also occur in the other direction, that is from no prediction to prediction.The determination section 10 may correspond to CPU 101 and / or Al processor 101 of Fig. 9. It is noted that any division into sections or units, e.g., determination section 10, including Al model 15, is only made for illustration purposes and that the present disclosure is not limited to any specific division of functions in specific units or sections.Error microphone 5 may also capture sound before noise cancellation. That is, error microphone 5 may be used as a second reference microphone for capturing an additional reference signal (e.g., second reference noise signal) based on sound between the outside sidewall la of electronic device 1 (e.g., headphone membrane) and ear 6.The second reference signal may be transmitted to the determination section 10. Thus, an additional prediction, i.e., of a set of second noise signals, and selection of a predicted second noise signal, which may include a comparison, may be conducted as described above but based on the second reference signal. Alternatively, the second reference signal may be the only reference signal on which the prediction for noise cancellation is based.Also, a selection of a predicted second noise signal may be based on the predicted ambient noise signal selected from the set based on reference ambient noise signal of microphone 3. That is, the predicted second noise signal associated with the predicted ambient noise signal may be selected from the set of second noise signals as a basis for generating the anti-noise signal which is emitted by loudspeaker 4.Fig. 3 illustrates an embodiment of a set of predicted ambient noise signals compared to a reference ambient noise signal. Fig. 3 includes a horizontal axis indicating time for all four signals of the figure. The four vertical axes indicate the respective amplitude of the four audio signals. xO indicates the amplitude of the reference ambient noise signal 2. xl, x2 and x3 indicate the amplitudes of the respective audio signals of the predicted set of ambient noise signals 11. At tl the reference ambient noise signal is acquired for prediction. Therefore, the predicted set of ambient noise signals 11 are based on the reference ambient noise signals 2 up to tl. All three signals of the set 11 predict the future of the reference ambient noise signal 2 beyond tl. After the prediction the signals 17a, 17b and 13 of the set of signals 11 are each compared to the reference ambient noise signal 2 for purposes of selecting the best match. While the prediction is processed the reference ambient noise signal at timepoint tl+n is acquired. Thus, the comparison occurs in regard to the comparison interval tc, from the first timepoint tl to the second timepoint tl+1. Predicted ambient noise signal 13 of the set of signals is selected as it is most similar to the reference ambient noise signal 2 in the comparison interval tc. By contrast, in the comparison interval tc the difference between signals 17a of set 11 to reference ambient noise signal 2 as well as the difference of signal 17b of set 11 to reference ambient noise signal 2 is greater. Therefore, the anti-noise signal (14, Fig. 2) for noise cancellation is based on predicted ambient noise signal 13 as also the prediction beyond tl+n is more likely to match the future reference ambient noise signal 2, which is also illustrated in Fig. 3. The anti -noise signal may be a 180°- phase inverted signal of predicted ambient noise signal 13, which also matches the 180°-phase inversion of reference ambient noise signal 2 (at least after error correction, i.e., compensation of any phase-offset) and therefore cancels ambient noise signal 2.Fig. 4 illustrates an embodiment of a set of predicted ambient noise signals compared to a reference ambient noise signal if the first timepoint of the reference signal used for prediction is the same as the second timepoint of the reference signal used for selection. Fig. 4 includes a horizontal axis indicating time for all four signals of the figure. The four vertical axes indicate the respective amplitude of the four audio signals. xO indicates the amplitude of the reference ambient noise signal 2. xl, x2 and x3 indicate the amplitude of the respective audio signals of the predicted set of ambient noise signals 11. At tl the reference ambient noise signal is acquired forprediction. Therefore, the predicted set of ambient noise signals 11 are based on the reference ambient noise signals 2 up to tl. All three signals of the set 11 predict the future of the reference ambient noise signal 2 beyond tl. Furthermore, all three signals of the set 11 also predict the reference ambient noise signal 2 up to timepoint tl. Compared to the to the signal up to timepoint tl of signals 17a, 17b and 13 of Fig. 3 the part of the signals 17a, 17b and 13 of Fig. 4 up to tl are also predicted based on the reference ambient noise signal 2. After the prediction the signals 17a, 17b and 13 of the set of signals 11 are each compared to the reference ambient noise signal 2 for purposes of selecting the best match.The comparison occurs in regard to the comparison interval tc, from the start of the signals to the first timepoint tl. Predicted ambient noise signal 13 of the set of signals is chosen as it is most similar to the reference ambient noise signal 2 in the comparison interval tc. By contrast, the difference between signals 17a of the set 11 to reference ambient noise signal 2 as well as the difference of signal 17b of the set 11 to reference ambient noise signal 2 is greater in the comparison interval tc. Therefore, the anti-noise signal (14, Fig. 2) for noise cancellation is based on predicted ambient noise signal 13 as also the prediction beyond tl is more likely to match the future reference ambient noise signal 2, which is also illustrated in Fig. 4. The antinoise signal may be a 180°-phase inverted signal of predicted ambient noise signal 13, which also matches the 180°-phase inversion of reference ambient noise signal 2 (at least after error correction, i.e., compensation of any phase-offset) and therefore cancels ambient noise signal 2.Fig. 5 illustrates noise cancellation of an embodiment based on a predicted ambient noise signal. The acquired reference ambient noise signal 2 is used as input for a prediction 10a, which predicts the set of ambient noise signals 11. Based on the set of ambient noise signals 11 and the reference ambient noise signal 2 the predicted ambient noise signal 13 of the set of ambient noise signals 11 which matches the reference ambient noise signal 2 the most is selected in a selection 10b. Based on the predicted ambient noise signal 11 noise cancellation is performed.The prediction 10a, selection 10b and noise cancellation 10c may be included in the determination section 10 of Fig. 2. However, it is noted that any division into sections or units is only made for illustration purposes and that the present disclosure is not limited to any specific division of functions in specific units or sections, e.g., the determination section 10.Fig. 6 illustrates noise cancellation of an embodiment based on a source-separated predicted ambient noise signal. The acquired reference ambient noise signal 2 is source separated lOd into three different source separated reference ambient noise signals 2a. In prediction 10a for each source separated ambient noise signal 2a a set of ambient noise signals 11 is predicted. Also, inselection 10b for each source separated predicted set of ambient noise signals the predicted ambient noise signal 13 is selected. That is, from each set of ambient noise signals 11, the predicted ambient noise signal 13 of the set 11 which matches the corresponding source separated reference ambient noise signal 2a the most is selected. In Fig. 6, three sets of ambient noise signals 11 are predicted and therefore for each set 11 one predicted ambient noise signal 13 is selected leading to three predicted ambient noise signals 13 separated by source.Depending on context profile 19, corresponding to the estimated context of the environment of the reference ambient noise signal, which may, for example, be location-based and / or timebased, the predicted ambient noise signals 13 of the different sources are merged lOe. In Fig. 6 only predicted ambient noise signal-Sourcel and predicted ambient noise signal-Source 2 are merged. The noise cancellation 10c is performed based on the merged signal.Alternatively, the context profile may influence at the level of the source separation lOd, determining based on which sources source separation is performed, or on the level of the prediction 10a, determining based on which source separated ambient noise signals 2a sets 11 are generated, i.e., which sets 11 are generated, or based on the level of the selection, determining from which sets I l a predicted ambient noise signal is selected for noise cancellation.The source separation lOd, prediction 10a, selection 10b, merging lOe and noise cancellation 10c may be included in the determination section 10 of Fig. 2.The source separation lOd may be part of the generative Al model 15 of Fig. 2. Thus, the Al model 15 may have not only learned to predict sets of ambient noise signals 11, but also to source separate the reference ambient noise signal for prediction 10a, for example based on the context profile 19.It is noted that any division into sections or units is only made for illustration purposes and that the present disclosure is not limited to any specific division of functions in specific units or sections, e.g., the determination section 10, Al model 15.Fig. 7 illustrates a network including a server as well as electronic devices of an embodiment. Network 50 connects different devices for example electronic device 1 indicating headphones, and smartphone 52, the headphones correspond to electronic device 1 of Figs. 1 and 2, smartphone 52 may also be an electronic device 1 according to some embodiments, (e.g., 1 of Fig. 9), but also more headphones or other electronic devices 1 may be connected to network 50. A user of smartphone 52 may transmit a context profile (e.g., 19 of Fig. 6), for example a location-based profile to the server 51, e.g., a cloud server, via network 50. The server may include a database of context profiles. Electronic device 1 may have access to the contextprofiles of server 51 via network 50. Electronic device 1 may acquire the database or part of the database of server 51 of one or more context profiles via the network 51. In case of the context profile being a location-based profile, the location-based profile may be activated in electronic device 1 (headphones) as soon as the electronic device 1 approaches the vicinity of the location the location-based profile refers to. The activation may include a transmission of the context profile to electronic device 1, for example from server 51 via network 50, or it may include an activation of an already previously stored database stored within electronic device 1. In this way, the location-based prediction may be crowd-sourced via the network 50 and via multiple users of devices 52 or similar devices and of electronic devices 1. Alternatively, the context profile may be transmitted to the electronic device 1 via network 50 from the smartphone 52, i.e., without involving server 51.Fig. 8 illustrates an embodiment of a method for noise cancellation. At 41 a reference ambient noise signal is obtained, for example, received from a microphone (e.g., reference microphone 3 of Figs. 1, 2) or from another entity which stores the reference ambient noise signal. At 42 a set of ambient noise signals for noise cancellation is predicted based on the acquired reference ambient noise signal at a first time point. At 43 a predicted ambient noise signal from the set of ambient noise signals is selected based on the acquired reference ambient noise signal at a second point of time. At 44 noise cancellation is performed.Fig. 9 illustrates an embodiment of an electronic device for noise cancellation. Electronic device 1 may correspond to electronic device 1 of Figs. 1, 2 and 7. The electronic device 1 may be implemented as headphones, for example of Fig. 1. The electronic device may also be implemented as earphones, as a vehicle, a terminal computer, a smartphone (e.g., 52 of Fig. 7), a tablet or other mobile device, such as a laptop or a wearable device, such as smart glasses, or the like. The electronic device 1 includes a CPU 101 as processor. Additionally, or alternatively, other computation hardware, such as GPU, TPU, DSP etc. may be used. The electronic device 1 further includes camera(s) 106, microphone(s) 107 and loudspeaker(s) 108 that are connected to the processor 101. The processor 101 may for example implement prediction, selection, for example including comparison, and / or generation of the anti-noise signal as well as control of emitting the anti-noise signal, error correction and / or switching as described with regard to Figs. 1 to 6. The microphone 107 may be configured to receive any kind of audio signal and may be reference microphone 3 of Figs. 1 and 2. Another microphone 107 may be configured to receive any kind of audio signal and may be error microphone 5 of Figs. 1 and 2. The loudspeaker 108 may be configured to emit any kind of audio signal and may be loudspeaker 4 of Figs. 1 and 2. The loudspeaker 108 may be configured to emit the anti-noise signal 14 of Fig. 2. Otherloudspeakers 108 or the same loudspeaker 108 emitting the anti -noise signal may be configured to play other audio signals, for example music or audio signals from a phone call or the like. The camera 106 may be one or more cameras, such as an RGB camera, and IR camera, a ToF camera, for example, an iToF or dTof, an event-based camera or the like.The electronic device 1 further includes a user interface 109 that is connected to the processor 101. This user interface 109 acts as a man-machine interface and enables a dialogue between a user (e.g., user of ear 6 of Fig. 2) and the electronic device 1. For example, a user may make configurations to the system using this user interface 109.The electronic device 1 further includes a Bluetooth interface 104, and a WLAN interface 105. These units 104, 105 act as I / O interfaces for data communication with external devices. The units 104, 105 may also be used for offloading the computation from the electronic device 1 to the network (e.g., network 50 of Fig. 7), such as described in Fig. 7, e.g., to a cloud server (51, Fig. 7) or to another device (e.g., 52, Fig. 7), such as a device close to the user. An ethemet interface may also be possible. For example, additional loudspeakers, e.g., loudspeaker 4 of Fig. 2, microphones, e.g., reference microphones 3 or error microphones 5 of Figs. 2 and 3, and cameras, with WLAN or Bluetooth connection may be coupled to the processor 101 via these interfaces 104 and 105. That is loudspeaker 4 of Fig. 2 may be coupled to the processor 101 via the interfaces 104 and 105.The electronic device 1 further includes a data storage 102 and a data memory 103 (here a RAM). The data memory 103 is arranged to temporarily store or cache data or computer instructions for processing by the processor 101, for example the predicted set of ambient noise signals 11 of Figs. 3, 4 and 5 or the reference ambient noise signal 2 of Figs. 3, 4 and 5. The data storage 102 is arranged as a long-term storage, e.g., of a context-based profile (e.g., 19 of Fig. 6), such as a location-based profile, or the predicted set of ambient noise signals 11 (Figs. 3 and 4) or the reference ambient noise signal 2 (Figs. 3 and 4), which may be obtained via the processor 101 that may implement noise cancellation.The connection between the processor 101 and the camera 106 may include a camera serial interface (CSI). The CSI is an interface between a camera 106 and a host processor 101. Thus, control signals and data from the processor 101 to the camera 106 as well as from the camera 106 to the processor 101 may be sent.Furthermore, the electronic device 101 includes an artificial intelligence (Al) processor 110. The Al processor 110 may include a graphics processing unit (GPU) and / or a tensor processing unit 20 (TPU). The Al processor 110 may be configured to execute an Al model (e.g., an artificialneural network), for example, the generative artificial intelligence model 15 of Fig. 2, for prediction of the set of ambient noise signals 2 of Figs. 3 to 6.It should be recognized that the embodiments describe methods with an exemplary ordering of method steps. The specific ordering of method steps is however given for illustrative purposes only and should not be construed as binding. Changes of the ordering of method steps may be apparent to the skilled person.Please note that any division into sections or units is only made for illustration purposes and that the present disclosure is not limited to any specific division of functions in specific units or sections. For instance, the determination section 10 could be implemented by a respective programmed processor, field programmable gate array (FPGA) and the like.A method for controlling an electronic device, such as electronic device 1 discussed above, is for example described under reference of Fig. 8. The method can also be implemented as a computer program causing a computer and / or a processor, such as processor 101 of Fig. 9 discussed above, to perform the method, when being carried out on the computer and / or processor. In some embodiments, also a non-transitory computer-readable recording medium is provided that stores therein a computer program product, which, when executed by a processor, such as the processor described above, causes the method described to be performed.All units and entities described in this specification and claimed in the appended claims can, if not stated otherwise, be implemented as integrated circuit logic, for example on a chip, and functionality provided by such units and entities can, if not stated otherwise, be implemented by software.In so far as the embodiments of the disclosure described above are implemented, at least in part, using software-controlled data processing apparatus, it will be appreciated that a computer program providing such software control and a transmission, storage or other medium by which such a computer program is provided are envisaged as aspects of the present disclosure.Note that the present technology can also be configured as described below.[1] An electronic device comprising circuitry for noise-cancellation configured to: receive a reference ambient noise signal (2); predict a set of ambient noise signals (11) for noise-cancellation based on the acquired reference ambient noise signal (2) at a first point of time (tl);select a predicted ambient noise signal (13) from the set of ambient noise signals (11), based on the acquired reference ambient noise signal (2) at a second point of time (tl, tl+n); and perform noise-cancellation based on the predicted ambient noise signal (13).[2] The electronic device of (1), wherein selecting the predicted ambient noise signal (13) is based on comparing the set of ambient noise signals (11) to the acquired ambient noise signal (2) at a second point of time (tl, tl+n).[3] The electronic device of (1) or (2), wherein selecting the predicted ambient noise signal (13) is based on selecting from the compared set of ambient noise signals (11) the signal if the difference to the acquired reference ambient noise signal (2) at a second point of time (tl, tl+n) is below a threshold.[4] The electronic device of any one of (1) to (3), wherein selecting the predicted ambient noise signal (13) is based on comparing a part of the set of ambient noise signals (11) corresponding to the acquired reference ambient noise signal (2) at a second point of time (tl, tl+n).[5] The electronic device of any one of (1) to (4), wherein the first point of time (tl) occurs at the same time or earlier than the second point of time (tl, tl+n).[6] The electronic device of any one of (1) to (5), wherein performing the noise-cancellation is based on generating and emitting an anti-noise signal (14) based on the predicted ambient noise signal (13).[7] The electronic device of any one of (1) to (6), wherein the circuitry is further configured to error measure an audio signal after the anti-noise signal (14) was emitted for the noisecancellation.[8] The electronic device of any one of (1) to (7), wherein the circuitry is further configured to switch from noise-cancellation based on the predicted ambient noise signal (13) to noisecancellation based on the reference ambient noise signal (2) based on the error measurement.[9] The electronic device of (8), wherein the switching occurs if the error measurement is above a predetermined threshold.

[0010] The electronic device of any one of (1) to (9), wherein the circuitry is further configured to perform error correction based on the error measurement.

[0011] The electronic device of any one of (1) to (10), wherein the circuitry is further configured to perform source separation (lOd) of the reference ambient noise signal (2), and whereinpredicting the set of ambient noise signals (11) is based on a source separated ambient noise signal (2a).

[0012] The electronic device of any one of (1) to (11), wherein the circuitry is further configured to merge (lOe) at least two predicted ambient noise signals (13) of different sources, which are based on source separated sets of ambient noise signals (11), and wherein the noise-cancellation is based on the merged predicted ambient noise signal.

[0013] The electronic device of any one of (1) to (12), wherein the circuitry is further configured to obtain a location-based noise profile (19), and wherein performing the prediction (10a) is based on the location-based noise profile (19).

[0014] The electronic device of (13), wherein obtaining the location-based noise profile (19) is based on crowdsourcing of multiple users.

[0015] The electronic device of any one of (1) to (14), wherein the circuitry is further configured to: receive a second reference noise signal associated with the reference ambient noise signal (2); and predict a set of second noise signals based on the acquired second reference noise signal; and wherein performing noise-cancellation based on the predicted ambient noise signal (13) includes selecting a predicted second noise signal of the set of second noise signals based on the predicted ambient noise signal and performing noise-cancellation based on the predicted second noise signal.

[0016] The electronic device of any one of (1) to (15), wherein performing the prediction is based on generative artificial intelligence (15).

[0017] A method for noise-cancellation comprising: obtaining a reference ambient noise signal (2); predicting a set of ambient noise signals (11) for noise-cancellation based on the acquired reference ambient noise signal (2) at a first point of time (tl); selecting a predicted ambient noise signal (13) from the set of ambient noise signals (11), based on the acquired reference ambient noise signal (2) at a second point of time (tl, tl+n); and performing noise-cancellation based on the predicted ambient noise signal (13).

[0018] The method of (17), wherein selecting the predicted ambient noise signal (13) is based on comparing the set of ambient noise signals (11) to the acquired ambient noise signal (2) at a second point of time (tl, tl+n).

[0019] The method of (17) or (18), wherein selecting the predicted ambient noise signal (13) is based on selecting from the compared set of ambient noise signals (11) the signal if the difference to the acquired reference ambient noise signal (2) at a second point of time (tl, tl+n) is below a threshold.

[0020] The method of any one of (17) to (19), wherein selecting the predicted ambient noise signal (13) is based on comparing a part of the set of ambient noise signals (11) corresponding to the acquired reference ambient noise signal (2) at a second point of time (tl, tl+n).

[0021] The method of any one of (17) to (20), wherein the first point of time (tl) occurs at the same time or earlier than the second point of time (tl, tl+n).

[0022] The method of any one of (17) to (21), wherein performing the noise-cancellation is based on generating and emitting an anti-noise signal (14) based on the predicted ambient noise signal (13).

[0023] The method of any one of (17) to (22), wherein the method further includes error measuring an audio signal after the anti-noise signal (14) was emitted for the noise-cancellation.

[0024] The method of any one of (17) to (23), wherein the method further includes switching from noise-cancellation based on the predicted ambient noise signal (13) to noise-cancellation based on the reference ambient noise signal (2) based on the error measurement.

[0025] The method of (24), wherein the switching occurs if the error measurement is above a predetermined threshold.

[0026] The method of any one of (17) to (25), wherein the method further includes performing error correction based on the error measurement.

[0027] The method of any one of (17) to (26), wherein the method further includes performing source separation (lOd) of the reference ambient noise signal (2), and wherein predicting the set of ambient noise signals (11) is based on a source separated ambient noise signal (2a).

[0028] The method of any one of (17) to (27), wherein the method further includes merging (lOe) at least two predicted ambient noise signals (13) of different sources, which are based on source separated sets of ambient noise signals (11), and wherein the noise-cancellation is based on the merged predicted ambient noise signal.

[0029] The method of any one of (17) to (28), wherein the method further includes obtaining a location-based noise profile (19), and wherein performing the prediction (10a) is based on the location-based noise profile (19).

[0030] The method of (29), wherein obtaining the location-based noise profile (19) is based on crowdsourcing of multiple users.

[0031] The method of any one of (17) to (30), wherein the method further includes: receiving a second reference noise signal associated with the reference ambient noise signal (2); and predicting a set of second noise signals based on the acquired second reference noise signal; and wherein performing noise-cancellation based on the predicted ambient noise signal (13) includes selecting a predicted second noise signal of the set of second noise signals based on the predicted ambient noise signal and performing noise-cancellation based on the predicted second noise signal.

[0032] The method of any one of (17) to (31), wherein performing the prediction is based on generative artificial intelligence (15).

[0033] A computer program comprising program code causing a computer to perform the method according to anyone of (17) to (32), when being carried out on a computer.

[0034] A non-transitory computer-readable recording medium that stores therein a computer program product, which, when executed by a processor, causes the method according to anyone of (17) to (32) to be performed.

Claims

CLAIMS1. An electronic device comprising circuitry for noise-cancellation configured to: receive a reference ambient noise signal; predict a set of ambient noise signals for noise-cancellation based on the acquired reference ambient noise signal at a first point of time; select a predicted ambient noise signal from the set of ambient noise signals, based on the acquired reference ambient noise signal at a second point of time; and perform noise-cancellation based on the predicted ambient noise signal.

2. The electronic device of claim 1, wherein selecting the predicted ambient noise signal is based on comparing the set of ambient noise signals to the acquired ambient noise signal at a second point of time.

3. The electronic device of claim 2, wherein selecting the predicted ambient noise signal is based on selecting from the compared set of ambient noise signals the signal if the difference to the acquired reference ambient noise signal at a second point of time is below a threshold.

4. The electronic device of claim 3, wherein selecting the predicted ambient noise signal is based on comparing a part of the set of ambient noise signals corresponding to the acquired reference ambient noise signal at a second point of time.

5. The electronic device of claim 1, wherein the first point of time occurs at the same time or earlier than the second point of time.

6. The electronic device of claim 1, wherein performing the noise-cancellation is based on generating and emitting an anti-noise signal based on the predicted ambient noise signal.

7. The electronic device of claim 6, wherein the circuitry is further configured to error measure an audio signal after the anti-noise signal was emitted for the noise-cancellation.

8. The electronic device of claim 7, wherein the circuitry is further configured to switch from noise-cancellation based on the predicted ambient noise signal to noise-cancellation based on the reference ambient noise signal based on the error measurement.

9. The electronic device of claim 8, wherein the switching occurs if the error measurement is above a predetermined threshold.

10. The electronic device of claim 7, wherein the circuitry is further configured to perform error correction based on the error measurement.

11. The electronic device of claim 1, wherein the circuitry is further configured to perform source separation of the reference ambient noise signal, and wherein predicting the set of ambient noise signals is based on a source separated ambient noise signal.

12. The electronic device of claim 11, wherein the circuitry is further configured to merge at least two predicted ambient noise signals of different sources, which are based on source separated sets of ambient noise signals, and wherein the noise-cancellation is based on the merged predicted ambient noise signal.

13. The electronic device of claim 1, wherein the circuitry is further configured to obtain a location-based noise profile, and wherein performing the prediction is based on the locationbased noise profile.

14. The electronic device of claim 13, wherein obtaining the location-based noise profile is based on crowdsourcing of multiple users.

15. The electronic device of claim 1, wherein the circuitry is further configured to: receive a second reference noise signal associated with the reference ambient noise signal; and predict a set of second noise signals based on the acquired second reference noise signal; and wherein performing noise-cancellation includes selecting a predicted second noise signal of the set of second noise signals based on the predicted ambient noise signal and performing noise-cancellation based on the predicted second noise signal.

16. The electronic device of claim 1, wherein performing the prediction is based on generative artificial intelligence.

17. A method for performing noise-cancellation comprising: obtaining a reference ambient noise signal; predicting a set of ambient noise signals for noise-cancellation based on the acquired reference ambient noise signal at a first point of time; selecting a predicted ambient noise signal from the set of ambient noise signals, based on the acquired reference ambient noise signal at a second point of time; and performing noise-cancellation based on the predicted ambient noise signal.

18. The method of claim 17, wherein selecting the predicted ambient noise signal is based on comparing the set of ambient noise signals to the acquired ambient noise signal at a second point of time.

19. The method of claim 18, wherein selecting the predicted ambient noise signal is based on selecting from the compared set of ambient noise signals the signal if the difference to the acquired reference ambient noise signal at a second point of time is below a threshold.

20. The method of claim 19, wherein selecting the predicted ambient noise signal is based on comparing a part of the set of ambient noise signals corresponding to the acquired reference ambient noise signal at a second point of time.

Images