Audio device, related method and system using codec information based processing

Abstract

The present disclosure relates to an audio device, related methods and systems using codec information-based processing. An audio device includes a processor, the processor includes a decoder and a first signal processor, wherein the audio device is configured to: obtain an audio input signal from a transmitter device, wherein the audio input signal is an encoded audio input signal encoded based on encoder parameters; decode the audio input signal using a decoder and decoder parameters to provide a decoder output signal; obtain codec information indicating encoder parameters and / or decoder parameters; and use the first signal processor and process the decoder output signal based on the codec information to provide a first signal processor output signal, wherein processing the decoder output signal using the first signal processor includes: using the first signal processor to process the decoder output signal by applying a generative model of the processing based on the codec information.

Description

Technical Field

[0001] The present disclosure relates to the field of audio devices and methods of operating audio devices, and in particular to audio devices and related methods utilizing codec information based processing. Background Art

[0002] In audio communications, more and more audio devices (such as wireless audio devices) compete to share the same available "talk time" and available effective bandwidth, which makes it a challenge to accommodate all audio devices in the available bandwidth. This may cause delays, dropped calls, and generally a higher probability of having a poor audio communication experience. In order to accommodate more audio devices simultaneously in the available bandwidth, the bit rate of the audio can be reduced. For example, the encoding and decoding of audio signals is a technique for reducing the size of audio signals by reducing the bit rate of the audio. However, the encoding and decoding process may reduce the quality of the audio signal, for example, due to lossy compression and / or undesirable characteristics, such as artifacts. It is challenging to encode and decode audio signals while maintaining the quality of the audio signal. This may be particularly a problem for audio device users in the communications department (such as users working in call centers) who use and / or wear their audio devices throughout the working day. By providing low-quality communications and end-user experience, it will cause users to lose focus and therefore reduce work efficiency. However, it is challenging to maintain high audio quality while using low or very low bit rates. Summary of the invention

[0003]

[0006] Therefore, there is a need for audio devices and methods for operating audio devices utilizing codec information based processing that can mitigate, alleviate or address existing shortcomings and can provide improved signal processing.

[0004] An audio device is disclosed. The audio device may be configured to be used as a receiver device. The audio device includes an interface, an audio speaker, and a microphone. The audio device includes one or more processors and a memory. The one or more processors include a decoder and a first signal processor. The first signal processor is configured to operate according to a generative model for processing based on a generative signal. The audio device is configured to obtain an audio input signal from a transmitter device, wherein the audio input signal is an encoded audio input signal encoded based on one or more encoder parameters. For example, the audio device is configured to decode the audio input signal using a decoder and one or more decoder parameters for providing a decoder output signal. The audio device is configured to obtain codec information indicating one or more encoder parameters and / or one or more decoder parameters. For example, the audio device is configured to use a first signal processor and process a decoder output signal based on the codec information for providing a first signal processor output signal. For example, processing the decoder output signal using the first signal processor includes: processing the decoder output signal using the first signal processor by applying a generative model based on the processing of the codec information. Using the first signal processor to process the decoder output signal by applying a generative model based on the processing of the codec information may include applying diffusion-based processing of the decoder output signal.

[0005] A method of operating an audio device configured to function as a receiver device, such as an audio device, is disclosed. The method includes obtaining an audio input signal from a transmitter device, wherein the audio input signal is an encoded audio input signal encoded based on one or more encoder parameters. For example, the method includes decoding the audio input signal using a decoder and one or more decoder parameters to provide a decoder output signal. The method includes obtaining codec information indicating one or more encoder parameters and / or one or more decoder parameters. For example, the method includes using a first signal processor and processing the decoder output signal based on the codec information to provide a first signal processor output signal, wherein, for example, processing the decoder output signal using the first signal processor includes: processing the decoder output signal using the first signal processor, for example, by applying a generative model of processing based on the codec information.

[0006] A computer-implemented method for training a generative model based on generative signal processing is disclosed. The method can be based on diffusion model-based training such as U-net, multi-resolution layers, and / or encoder-decoder architectures. The method includes obtaining an audio data set including one or more audio signals. The method includes generating a codec-distorted audio data set by encoding one or more audio signals encoded, for example, based on one or more encoder parameters to provide one or more encoded audio signals. For example, the method includes generating a codec-distorted audio data set by decoding one or more encoded audio signals using a decoder and one or more decoder parameters to provide one or more codec-distorted audio signals. The method includes combining one or more audio signals with one or more white noise signals to provide a white noise audio data set including one or more white noise audio signals. For example, the method includes determining one or more estimated white noise signals and one or more estimated audio signals by applying a generative model to one or more white noise audio signals and one or more codec-distorted audio signals. The method includes training a generative model based on one or more audio signals, one or more estimated audio signals, one or more estimated white noise signals, and one or more white noise signals.

[0007] The disclosed system, audio device, and related method provide improved signal processing of input signals including audio. In other words, the present disclosure can provide improved processing based on codec information of input signals including audio. It is understood that the present disclosure provides notification signal processing of input signals including audio. For example, the present disclosure can provide notification enhancement of voice signals, for example, based on codec information. It is understood that the present disclosure can provide voice enhancement of input signals including audio. The present disclosure can maintain high audio signal quality while using low or very low bit rates. This in turn can allow a large number of audio devices to be on limited bandwidth while maintaining high audio communication quality.

[0008] An advantage of the present disclosure is the use of generative methods, such as generative models, that avoid the limitation of only adjusting the original signal as in mask-based methods. Instead, generative methods allow for the generation of new signals.

[0009] Advantageously, the disclosed method may enable the quality of a decoded audio signal to be improved, such as by reducing undesirable characteristics of the audio signal, such as artifacts that may have been generated during encoding and / or decoding.

[0010] It is to be understood that the present disclosure can provide improved signal processing, such as improved tuning of a signal processing pipeline. Signal processing can be improved based on one or more encoder parameters and / or one or more decoder parameters. Subsequently, the present disclosure can provide an improved output signal, such as an improved audio output quality at the receiver end, for a receiver end user. In other words, for example, the present disclosure can provide improved audio quality at the receiver end during a call or conference.

[0011] The advantage of the present disclosure is that an improved communication experience, such as voice communication, is achieved between a sender side (such as a sender audio device) and a receiver side (such as a receiver audio device). For example, the present disclosure can provide an improved communication experience for telephone and / or telephone conferences using, for example, headphones, speaker phones, and / or video bars. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The above and other features and advantages of the present disclosure will become apparent to those skilled in the art through the following detailed description of examples of the present disclosure with reference to the accompanying drawings, in which:

[0013] Figure 1 schematically illustrates an example system according to the present disclosure,

[0014] FIG. 2A to FIG. 2B is a flowchart of an example method according to the present disclosure, and

[0015] Figure 3 is a flow chart of an example method according to the present disclosure. DETAILED DESCRIPTION

[0016] Various examples and details will be described below with reference to the relevant drawings. It should be noted that the drawings may or may not be drawn to scale, and in all drawings, elements of similar structure or function are represented by similar reference numerals. It should also be noted that the drawings are intended only to facilitate the description of the examples. They are not intended to be an exhaustive description of the present disclosure or to limit the scope of the present disclosure. In addition, the examples shown do not need to have all the aspects or advantages shown. The aspects or advantages described in conjunction with a specific example are not necessarily limited to that example and may be practiced in any other example, even if not shown or not so explicitly described.

[0017] For the sake of clarity, the drawings are schematic and simplified, and they only show details that are helpful for understanding the present disclosure, while other details have been omitted. Throughout, the same reference numerals are used for the same or corresponding parts.

[0018] An audio device is disclosed. The audio device is configured to be used as a receiver device. In other words, the audio device is configured to receive an input signal from an audio device configured to be used as a transmitter device. The audio device as disclosed herein may include an interface, an audio speaker, one or more microphones, one or more processors, and a memory. The one or more processors include a decoder and a first signal processor.

[0019] Decoders as disclosed herein can be implemented as hardware and / or software on audio equipment.Decoders as disclosed herein can be considered as being configured to convert and / or decode the input signal from a transmitter device into a module of a decoder output signal.In other words, the decoder can be configured to decompress the input signal from the transmitter device.For example, the decoder can be configured to convert and / or decode the input signal from the transmitter device into an audio signal.

[0020] An audio device may be considered as an audio device configured to obtain an input signal (such as an audio signal), output an audio signal, and process an input signal (such as an audio signal). An audio device may be considered as or include a headset, a speakerphone, and / or an audio / video integrated machine. For example, an audio device may be considered as a conference audio device, for example, a conference audio device is configured to be used by one party (such as one or more users at the near end) to communicate with one or more other parties (such as one or more users at the far end). When an output signal is sent back to the far end, an audio device configured to be used as a receiver device may also be configured to be used as a transmitter device. Therefore, a receiver audio device and a transmitter audio device may be switched between being a receiver audio device and being a transmitter audio device. An audio device may be considered as an intelligent audio device. An audio device may be used for conferences and / or meetings between two or more parties that are far away from each other. An audio device may be used by one or more users near the location where the audio device is located (also referred to as the near end). An audio device may be configured to output an audio device output at a receiver end, such as using an audio speaker and based on an input signal. The audio device output may be considered to be an audio output signal which is the output of an audio speaker at the proximal end of the location where the audio device and the user of the audio device are located.

[0021] In one or more examples, the audio device can be considered as an enhancer device, such as an audio enhancer device. The enhancer device can be considered as a device for enhancing the audio quality of the encoded / decoded audio signal for providing an enhanced audio output. For example, the enhancer device can enhance one or more audio parameters of the encoded / decoded audio signal based on the codec information for providing an enhanced audio output.

[0022] An audio device may be a single audio device. An audio device may be considered as a plurality of interconnected audio devices, such as a system, for example, an audio device system. The system may include one or more users. It should be noted that the term speaker may be considered as a user of an audio device.

[0023] In one or more example audio devices, the interface includes a wireless transceiver (also represented as a radio transceiver) and an antenna for wirelessly transmitting and receiving input signals (such as audio signals), such as for wireless transmission of output signals and / or wireless reception of wireless input signals. The audio device can be configured for wireless communication with one or more electronic devices (such as another audio device, a smart phone, a tablet computer, a computer, and / or a smart watch). The audio device optionally includes an antenna for converting one or more wireless input audio signals into antenna output signals. The audio device system and / or the audio device can be configured for wireless communication via a wireless communication system (such as a short-range wireless communication system, such as Wi-Fi, Bluetooth, Zigbee, IEEE 802.11, IEEE 802.15, infrared, etc.).

[0024] The audio device system and / or the audio device may be configured for wireless communication via a wireless communication system, such as a 3GPP system, such as a 3GPP system supporting one or more of New Radio (NR), Narrowband IoT (NB-IoT) and Long Term Evolution Enhanced Machine Type Communication (LTE-M), millimeter wave communication (such as millimeter wave communication in a licensed band, such as device-to-device millimeter wave communication in a licensed band).

[0025] In one or more example audio device systems and / or audio devices, the interface of the audio device includes one or more of a Bluetooth interface, a Bluetooth low energy interface, and a magnetic induction interface. For example, the interface of the audio device may include a Bluetooth antenna and / or a magnetic interference antenna.

[0026] The audio device system may be considered an audio communication system. The system may be considered a communication system for performing calls (such as audio and / or video calls). A system such as a communication system may be considered a system utilizing processing based on codec information. The audio device system may include an audio device, such as an audio device configured to function as a receiver device as disclosed herein, and a transmitter audio device, such as an audio device configured to function as a transmitter device as disclosed herein.

[0027] In one or more example audio devices, the interface may include a connector for wired communication via a connector (such as by using a cable). The connector may connect one or more microphones to the audio device. For example, the connector may connect the audio device to an electronic device for a wired connection. The connector may be viewed as an electrical connector, such as a physical connector for connecting an audio device to another device via a wire.

[0028] One or more interfaces may be or include a wireless interface, such as a transmitter and / or a receiver, and / or a wired interface, such as a connector for physical coupling. For example, an audio device may have an input interface configured to receive data (such as a microphone input signal). In one or more example audio devices, the audio device may be used in all form factors in all types of environments, such as for headphones and / or video conferencing equipment. For example, an audio device may not have a specific microphone placement requirement. In one or more example audio devices, the audio device may include an external microphone.

[0029] An audio device configured to be used as a receiver device is disclosed. The audio device includes an interface, an audio speaker, and a microphone, the audio device includes one or more processors and a memory, the one or more processors include a decoder and a first signal processor. The first signal processor is configured to operate according to a generative model based on processing of a generative signal.

[0030] The audio device is configured to obtain an audio input signal from a transmitter device, wherein the audio input signal is an encoded audio input signal encoded based on one or more encoder parameters.

[0031] For example, the audio device is configured to obtain (such as receive and / or retrieve) an audio input signal via a wired and / or wireless connection to a transmitter device. For example, the input audio signal can be encoded by the transmitter device. For example, the transmitter device is an electronic device external to the disclosed audio device, such as another audio device at a remote end. It is understood that the transmitter device can be an electronic device, such as an electronic device including an audio device. For example, the transmitter device can include a smart phone and / or a computer, such as a laptop computer.

[0032] A sender device may be considered as a device configured to send an audio signal to an audio device configured to function as a receiver device. For example, an audio output signal provided by a sender device may be considered as an audio input signal of a receiver device.

[0033] In some examples, an audio device may include a transmitter device, eg, a transmitter device configured to provide (eg, transmit) an audio input signal via one or more internal connections of the audio device.

[0034] In one or more example audio devices, the audio device is configured to obtain an audio input signal from a remote end (such as a remote party or user). In other words, the transmitter device can be considered as an audio device at the remote end. For example, the audio input signal includes audio. For example, the audio input signal has been encoded at the transmitter device, for example, based on one or more encoder parameters.

[0035] In one or more embodiments or examples, for example, the audio input signal undergoes signal processing, such as compression and / or enhancement, at the transmitter device. The audio input signal may indicate an audio signal generated by a user at a far end. In other words, the audio input signal may indicate speech, such as speech from a far end transmitter device. The audio input signal may be based on and / or considered as an output signal of a transmitter device, such as an output signal of a signal processor of the transmitter device.

[0036] The first signal processor may be configured to operate according to a generative model based on one or more diffusion machine learning techniques. For example, the generative model may be viewed as a generative model for processing based on a generative signal. For example, the generative model may be viewed as a generative model for processing based on codec information. For example, the generative model may be configured to perform denoising, such as deep noise reduction and / or echo cancellation, for example, using one or more diffusion machine learning techniques.

[0037] In one or more example audio devices, the first signal processor includes a machine learning circuit configured to operate according to a machine learning model, such as a generative model.

[0038] For example, the machine learning model may include or utilize a neural network, artificial intelligence, deep learning, and / or machine learning. The machine learning model may be viewed as a model for signal processing. For example, the machine learning model may be viewed as a model for decoder output signal processing based on codec information.

[0039] In one or more example audio devices and / or audio device systems, the machine learning model includes a model layer, which includes an input layer, one or more intermediate layers, and an output layer for processing a decoder output signal (such as an input signal). In one or more example audio devices and / or audio device systems, the machine learning model includes a neural network. In one or more example audio devices and / or audio device systems, the machine learning model includes a neural network layer, which includes an input layer, one or more intermediate layers, and an output layer for processing a decoder output signal (such as an input signal). In other words, the input layer, one or more intermediate layers, and / or the output layer can be regarded as layers of the machine learning model, such as layers of a neural network. One or more intermediate layers can be regarded as hidden layers (such as hidden features). One or more intermediate layers can include a first intermediate layer.

[0040] Models as referred to herein (such as generative models and / or non-generative models) can be regarded as models and / or schemes and / or mechanisms and / or methods configured to process one or more input signals (such as decoder output signals and / or signal processor input signals) based on codec information and / or previous models.

[0041] In one or more example audio devices and / or audio device systems, a generative model as mentioned herein may be stored on a non-transient storage medium (e.g., on a memory of an audio device). A generative model may be stored on a non-transient storage medium of an audio device configured to execute the model. In one or more example audio devices and / or audio device systems, a generative model may include model data and / or computer-readable instructions (e.g., based on codec information). The generative model data and / or computer-readable instructions may be used by an audio device. A generative model (such as model data and / or computer-readable instructions) may be used by an audio device to process a decoder output signal, such as an audio input signal.

[0042] Generative signal-based processing can be viewed as signal processing that utilizes generative models to analyze, manipulate, and / or synthesize signals (such as audio signals). Generative models are a type of artificial intelligence model that can learn patterns and structures from existing data and then generate new data that follows similar patterns. These models are trained on data sets containing examples of the signal types of interest. Once trained, generative models can be used for a variety of tasks, such as signal denoising, enhancement, and generating new types of signals. In the present disclosure, an audio device is configured to apply generative signal-based processing to a decoder output signal based on codec information (such as using codec information for adjustment).

[0043] Generative signal-based processing may include the use of generative adversarial networks (GANs), variational autoencoders (VAEs), and / or diffusion models (such as U-net, multi-resolution layers, and / or encoder-decoder architectures).

[0044] For example, the audio input signal includes one or more signals indicating audio. In one or more embodiments or examples, the audio input signal undergoes signal processing, such as encoding, compression and / or enhancement, at the transmitter device. The audio input signal can indicate an audio signal generated by a user at a far end. In other words, the audio input signal can indicate a voice, such as a voice from a far-end transmitter device, and / or other sounds generated at a far end. The audio input signal can be based on and / or be considered as an output signal of a transmitter device, such as an output signal of a signal processor of the transmitter device. For example, the audio input signal can be considered as an encoded audio input signal encoded at the transmitter device. In other words, the audio input signal can be considered as an audio input encoded by the transmitter device before being sent to a receiver device. For example, the audio device is configured to obtain (such as receive and / or retrieve) an encoded audio input signal.

[0045] Encoder parameters (such as one or more encoder parameters) may be considered as parameters related to the encoding or encoding technique of the audio input signal, e.g., for providing the encoded audio input signal. In other words, the encoder parameters may be parameters with which the audio input signal has been encoded, e.g., for providing the encoded audio input signal. The encoder parameters may include one or more of a sampling rate parameter, a bit depth parameter, a channel configuration parameter, a frame size parameter, a filtering parameter, a preprocessing parameter, an audio decoding format parameter, a psychoacoustic model parameter, a predictor selection parameter, an entropy decoding parameter, and a variable bit rate mode parameter.

[0046] The audio device is configured to decode the audio input signal using the decoder and one or more decoder parameters to provide a decoder output signal.It is understood that the decoder can be configured to decode, extract, and / or obtain one or more audio signals from the audio input signal.

[0047] Decoder parameters (such as one or more decoder parameters) may be considered as parameters related to the decoding or decoding technique of the audio input signal, for example, for providing an encoded audio input signal. In other words, the decoder parameters may be parameters with which the audio input signal has been decoded, for example, for providing a decoder output signal. The decoder parameters may include one or more of sampling rate conversion parameters, channel mapping parameters, downmix parameters, bass management parameters, equalization setting parameters, reverberation effect parameters, pitch stretching parameters, time stretching parameters, jitter parameters, and normalization parameters.

[0048] For example, the decoder output signal can be regarded as the first signal processor input signal. In one or more examples, the decoder output signal can be regarded as the decoded audio input signal. For example, the decoder output signal includes codec information. In some examples, for example, the codec information can be extracted from the decoder output signal by the decoder.

[0049] In some examples, the audio device may be configured to decode the audio input signal using a decoder and one or more decoder parameters for providing codec information.

[0050] In some examples, for example, codec information may be provided (eg, transmitted) to an audio device via a transmitter device.

[0051] The audio device is configured to obtain codec information indicating one or more encoder parameters and / or one or more decoder parameters. The codec information can be regarded as information associated with one or more encoder parameters and / or one or more decoder parameters. In other words, the codec information can include one or more encoder parameters and / or one or more decoder parameters. Alternatively or additionally, the codec information can be determined based on one or more encoder parameters and / or one or more decoder parameters. For example, the codec information can be regarded as information describing which codec and / or which codec parameter has been used to encode and / or decode the audio input signal.

[0052] It will be appreciated that codec information may be included in a decoder output signal, extracted by a decoder (such as when decoding an audio input signal and / or after decoding an audio input signal), extracted by a codec modulator as disclosed herein, and / or provided as auxiliary information from a transmitter device (e.g., sent by a transmitter device to a receiver device as auxiliary information of an audio input signal).

[0053] The audio device is configured to process the decoder output signal using the first signal processor and based on the codec information for providing a first signal processor output signal.

[0054] In some examples, processing the decoder output signal to provide the first signal processor output signal includes, for example, generating the first signal processor output signal based on the decoder output signal, for example, using the codec information as a codec modifier. In one or more example audio devices, the decoder output signal can be considered as the first signal processor input signal.

[0055] In some examples, the audio device is configured to use a first signal processor and process a decoder output signal based on the codec information to provide an audio output signal. For example, the first signal processor output signal may include an audio output signal. In other words, processing the decoder output signal may include generating a new enhanced signal, e.g., the first signal processor output signal, based on the decoder output signal and using the codec information as an adjustment in processing the decoder output signal.

[0056] It should be understood that the expression "based on" as disclosed herein may be viewed or understood as: "according to", "as input to", and / or "depending on". For example, processing the decoder output signal based on codec information may be understood as processing the decoder output signal according to the codec information, using the codec information as input in processing the decoder output signal, and / or depending on the codec information.

[0057] In one or more example audio devices, the codec information includes one or more of a codec type, a sampling rate, and a bit rate.

[0058] For example, the codec type may be considered to indicate the codec type according to which the audio input signal is encoded. For example, the audio input signal may be encoded according to one or more codecs, such as a codec type (e.g., EVS, MP3, LC3, MP4A, FLAC, etc.).

[0059] For example, the sampling rate can be thought of as the number of samples collected per unit time.

[0060] For example, bit rate can be thought of as the amount of data encoded per unit of time.

[0061] In one or more example audio devices, processing the decoder output signal using the first signal processor includes processing the decoder output signal based on one or more of a codec type, a sampling rate, and a bit rate.

[0062] For example, processing the decoder output signal with the first signal processor includes processing the decoder output signal with the first signal processor by applying a generative model of processing based on codec information.

[0063] For example, the generative model may be configured to process the decoder output signal based on codec information. For example, the generative model may be configured to process the decoder output signal based on a codec according to which the decoder output signal has been previously encoded and decoded. For example, when the decoder audio signal has been encoded and decoded according to a given codec (e.g., LP3), the decoder output signal may include one or more undesirable features associated with the codec (e.g., audio artifacts, defects, etc.). For example, during encoding and / or decoding of an audio signal (such as a decoder output signal), one or more audio artifacts and / or defects may be generated. In some examples, processing the decoder output signal by applying a generative model based on processing of codec information may enable the audio output signal to be processed, such as improving the quality of the first signal processor output signal, such as by processing the decoder output signal in a manner corresponding to the encoding and / or decoding technique applied to the decoder output signal. This may advantageously enable undesirable features (e.g., audio artifacts and / or defects) of the decoder output signal to be processed with improved accuracy and / or robustness (e.g., to improve the quality of the first signal processor output signal (e.g., audio device output signal)).

[0064] It can be understood that by directing the processing implemented by the first signal processor, the audio device can reduce the introduction of noise and artifacts in the processing of the input signal at the receiver device. In other words, by directing the processing implemented by the first signal processor, the audio device can reduce the introduction of noise and artifacts in the receiver output signal.

[0065] In one or more example audio devices, the one or more processors include an audio conditioner configured to process a decoder output signal for providing one or more audio features. In one or more example audio devices, processing the decoder output signal using the first signal processor includes: processing the decoder output signal based on the one or more audio features.

[0066] The audio regulator can be regarded as a regulator for regulating the processing of the decoder output signal, such as regulating the processing of the decoder output signal based on one or more audio features. In other words, the audio device can be configured to use one or more audio features provided by the audio regulator as input, or adjust the processing of the decoder output signal by the first signal processor. The audio regulator can be configured to extract one or more audio features from the decoder output signal. In some examples, for example, the audio regulator can be regarded as a distorted audio regulator configured to identify the distorted audio in the decoder output signal. In some examples, the audio regulator can be regarded as a voice regulator. In some examples, for example, the audio regulator can be a distorted voice regulator configured to identify the distorted voice in the decoder output signal. In one or more examples or embodiments, the audio regulator can be regarded as a pre-processing module for processing the decoder output signal. For example, the audio regulator can pre-process the decoder output signal for providing one or more audio features from the decoder output signal, and these audio features can be provided as input or provided to adjust the processing of the decoder output signal by the first signal processor.

[0067] For example, the audio conditioner includes a neural network, such as a convolutional neural network (CNN), an autoregressive neural network, and / or a recurrent neural network (RNN).

[0068] In one or more examples, the audio device can be configured to process the decoder output signal using one or more audio features as input and / or conditioning (eg, using a first signal processor).

[0069] The one or more audio features may be considered as features indicative of one or more audio characteristics of the decoder output signal.

[0070] In one or more example audio devices, one or more processors include a codec regulator configured to determine codec information. It is understood that the codec regulator may be provided with a payload of an audio input signal and / or a payload of a decoder output signal as input to determine the codec information. In one or more example audio devices, obtaining the codec information includes: determining the codec information using the codec regulator, and providing the codec information to the first signal processor.

[0071] The codec regulator can be viewed as a regulator for regulating the processing of the decoder output signal, such as regulating the processing of the decoder output signal based on the codec information. In other words, the audio device can be configured to use the codec information provided by the codec regulator as input, or to regulate the processing of the decoder output signal by the first signal processor. The codec regulator can be configured to extract one or more audio features from the decoder output signal.

[0072] For example, the codec regulator is configured to determine the codec information using a neural network. In some examples, for example, the codec information can be extracted by the codec regulator from the decoder output signal.

[0073] In one or more example audio devices, obtaining codec information includes obtaining codec information from a transmitter device.

[0074] It is to be understood that the codec information may be provided by the transmitter device, for example, included in the audio input signal. For example, the codec information may be provided in a header packet of the audio input signal and / or provided along with the audio input signal when the audio input signal is received from the transmitter device. In other words, the codec information may be obtained via auxiliary information included in or accompanying the audio input signal. For example, the codec information may be extracted from the header packet by the codec adjuster.

[0075] In one or more example audio devices, one or more processors include a second signal processor, which is configured to operate according to a non-generative model based on a process of a non-generative signal. In one or more example audio devices, the second signal processor is configured to process a second signal processor input signal for providing a second signal processor output signal, wherein the second signal processor input signal is based on a decoder output signal. In one or more example audio devices, the audio device is configured to use the second signal processor to process the decoder output signal. It is understood that the second signal processor can operate in parallel with the first signal processor. The second signal processor input signal can be based on the decoder output signal. For example, the second signal processor can be configured to perform mask-based processing on the decoder output signal.

[0076] In one or more example audio devices, the second signal processor includes a noise reducer and / or an echo controller. The echo controller can be viewed as a module configured to perform linear echo cancellation and / or residual echo cancellation. The second signal processor can be viewed as and / or represents a deep noise reducer, such as a module using deep noise reduction (DNS). It is understood that the second signal processor can be configured to generate a time-frequency, for example, to enhance one or more characteristics of the input signal, such as an enhanced decoder output signal. The second signal processor can be configured to operate according to a non-generative model, which includes two gated recurrent units (GRUs) and a fully connected dense layer. The second signal processor can be configured to improve signal processing in a wideband mode (WB), an ultra-wideband mode (SWB), and / or a full-band mode (FB).

[0077] In one or more example audio devices, for example, processing the second signal processor input signal includes using the second signal processor and / or based on codec information to reduce noise of the second signal processor input signal, such as reducing noise of a decoder output signal. In other words, processing the second signal processor input signal includes using the second signal processor and / or based on codec information to reduce noise of the second signal processor input signal for providing a second signal processor output signal with reduced noise. For example, processing the second signal processor input signal includes using the second signal processor and / or based on codec information to denoise the second signal processor input signal, for example, for providing a denoised first signal processor output signal.

[0078] In one or more example audio devices, the second signal processor includes a machine learning circuit configured to operate according to a machine learning model (such as a non-generative model). It will be appreciated that the second signal processor can be configured to operate according to a machine learning model (such as a non-generative model) for performing mask-based processing, such as using a mask-based technique.

[0079] For example, the non-generative model can be configured to perform processing based on the non-generative signal. In one or more example methods, the non-generative model can be configured to implement a mask-based method. In other words, the processing based on the non-generative signal can include a mask-based method. In one or more example audio devices, reducing the noise of the second signal processor input signal includes reducing the noise of the second signal processor input signal according to a mask-based method.

[0080] In one or more example audio devices, the audio device is configured to determine signal-to-noise ratio information based on a decoder output signal. In other words, the audio device can be configured to determine information indicating a signal-to-noise ratio (SNR) based on the decoder output signal. The SNR information can be considered as information indicating the SNR of the decoder output signal.

[0081] Signal-to-noise ratio (SNR) can be considered a quantitative measurement for assessing signal quality relative to the presence of unwanted or irrelevant background noise. SNR can represent the ratio of the power or amplitude of a signal (the "signal" component) to the power or amplitude of the accompanying noise (the "noise" component). Higher SNR values ​​can indicate a stronger, more discernible signal compared to the noise, resulting in better overall signal quality and more accurate data interpretation. In practice, a higher SNR generally means a clearer and more reliable signal that is less likely to be corrupted by unwanted jammers or interference. It will be appreciated that SNR information can be determined and / or extracted by a decoder from an audio input signal.

[0082] In one or more example audio devices, processing the decoder output signal includes: processing the decoder output signal based on the signal-to-noise ratio information. For example, the audio device can be configured to use the first signal processor and / or the second signal processor to process the decoder output signal based on the signal-to-noise ratio information and the codec information.

[0083] In one or more example audio devices, the audio device includes a mixer configured to combine a first signal processor output signal and a second signal processor output signal for providing an audio output signal.

[0084] In one or more example audio devices, a first signal processor output signal and a second signal processor output signal may be combined to generate and / or determine an audio device output. In one or more audio devices, an audio device (such as a first signal processor) includes a mixer, and the mixer is configured to combine the first signal processor output signal and the second signal processor output signal to provide the first signal processor output signal. In this case, the first signal processor output signal may be considered as a mixer output signal. For example, the mixer may be configured to add the first signal processor output signal to the second signal processor output signal to provide the first signal processor output signal. For example, the mixer may be configured to generate a mixer output signal including a portion of the first signal processor output signal and a portion of the second signal processor output signal.

[0085] In one or more example audio devices, combining the first signal processor output signal and the second signal processor output signal is based on signal-to-noise ratio information. By combining the first signal processor output signal and the second signal processor output signal based on SNR information, different parts of the first signal processor output signal and / or different parts of the second signal processor output signal can be combined for different regions of the audio input signal. For example, the first signal processor can be used to process the region of the audio input signal with a higher SNR, and the second signal processor can be used to process the region of the audio input signal with a lower SNR. Then, the mixer can combine the output of the processing of different regions by the first signal processor output signal and the second signal processor output signal. After being customized differently, the audio device can be configured to combine conventional mask-based enhancement methods, and based on the SNR calculated from the audio input signal, the first signal processor applying the generative model and the second signal processor using the mask-based processing are adjusted. This may be beneficial because it is well known that mask-based methods are limited in their achievable performance and saturate when the SNR is very low or very high, so they are limited in their performance. Based on the input SNR information at the time-frequency bins, a conditional estimator, such as a mixer, can be formulated that combines the mask-based output and the generated speech output from the enhancer.

[0086] In one or more examples or embodiments, combining the first signal processor output signal and the second signal processor output signal is based on an uncertainty parameter. For example, the audio device is configured to determine an uncertainty parameter indicating the uncertainty of the processing quality. In other words, the audio device can be configured to determine an uncertainty parameter indicating the uncertainty of the processing quality of the decoder output signal estimated by the first signal processor for providing the first signal processor output signal. The uncertainty parameter can indicate an estimate of the quality of the processing result of the decoder output signal, such as an estimate of the quality of the first signal processor output signal. After being customized differently, the uncertainty parameter can indicate the uncertainty of the prediction of the decoder output signal processed by the first signal processor as disclosed herein. For example, the uncertainty parameter can indicate the uncertainty of the prediction of the decoder output signal processed by the first signal processor as disclosed herein using a generative model. The uncertainty parameter can be regarded as an estimate of the confidence of the audio device to the processing quality when using the first signal processor (such as a generative model) for a given input (e.g., for a given decoder output signal). Nevertheless, an uncertainty parameter indicating a high uncertainty of the processing quality can lead to a good processing quality of the decoder output signal, and vice versa, an uncertainty parameter indicating a low uncertainty of the processing quality can lead to a poor processing quality of the decoder output signal.

[0087] By combining the first signal processor output signal and the second signal processor output signal based on the uncertainty parameter, the audio output can be adjusted based on the first signal processor output signal and / or the second signal processor output signal according to the uncertainty of the processing quality of the first signal processor. For example, when the uncertainty parameter meets a certain quality standard and / or threshold, the audio device can be configured to use the first signal processor output signal to provide audio output. On the one hand, when the uncertainty parameter does not meet the quality standard and / or threshold, the audio device can be configured to use the second signal processor output signal to provide audio output.

[0088] The mixer can then combine and / or alternate the first signal processor output signal and the second signal processor output signal for providing an audio output. Variously customized, the audio device can be configured to combine conventional mask-based enhancement methods and adjust the first signal processor that applies the generative model and the second signal processor that uses mask-based processing based on the uncertainty parameter.

[0089] In one or more example audio devices, the audio device is configured to determine a processing scheme for processing a decoder output signal. A processing scheme can be viewed as a structured plan or set of steps that summarizes or defines how to process a decoder output signal. In other words, a processing scheme can define a clear framework for applying various processing techniques to a decoder output signal. The steps and / or settings of the processing scheme can be defined or determined during training of a generative model as disclosed herein. The processing scheme can be determined by a scheduler during training of a generative model.

[0090] In one or more examples or embodiments, the processing scheme includes the number of iterations of processing and / or the noise reduction parameters associated with each iteration of processing. The processing scheme may indicate how many iterations of the decoder output signal processing are required to achieve a specific audio quality of the audio output signal. The processing scheme may determine how many iterations are required, and the noise reduction parameters may indicate how much noise is to be removed in each iteration of processing. The number of iterations and the noise reduction parameters may be fixed numbers or values ​​determined during the training of the generative model. In each processing iteration of the decoder output signal, the first signal processor may reduce the noise of the decoder output signal by a certain amount according to the noise reduction parameters. For example, the number of iterations may be the lowest number of iterations possible when the decoder output signal (such as the processed decoder output signal) meets the quality threshold. For example, the quality threshold may be determined based on a perceptual quality estimator (such as a non-intrusive quality measurement technique, such as DNS MOS and / or Nisqa) for speech and audio.

[0091] In one or more examples or embodiments, processing the decoder output signal using the first signal processor includes: processing the decoder output signal based on a processing scheme. In other words, processing the decoder output signal may include processing the decoder output signal using a certain number of iterations, such as using a generative model processing iteration. For example, processing the decoder output signal may include processing the decoder output signal using a certain number of processing iterations, each processing iteration having a specific noise reduction parameter.

[0092] In one or more example audio devices, the first signal processor includes a neural network that is a multi-resolution network. For example, the multi-resolution network can be viewed as the architecture of a neural network. A multi-resolution network is a type of neural network architecture designed to process and analyze data at multiple levels of resolution or scale. This approach is particularly useful when processing complex data that exhibits structures or patterns at different scales, such as audio signals, and specifically audio signals including codec artifacts and noisy audio signals. Multi-resolution networks enable the network to capture both fine-grained and coarse-grained information simultaneously, enhancing the network's ability to understand and represent the underlying features of the data. For example, a multi-resolution network can extract features at different time and frequency scales, enhancing the network's ability to understand complex time patterns. In other words, the first signal processor is capable of extracting features at different time and frequency scales. In one or more examples, the multi-resolution network is a network like a multi-resolution U-net.

[0093] In one or more examples or embodiments, the first signal processor includes a signal processing encoder and a signal processing decoder. The signal processing encoder can be regarded as a discriminator network. The signal processing decoder can be regarded as a generator network. It is understood that the signal processing encoder can be used to map input data to a low-dimensional latent space, and the signal processing decoder can be used to generate new data from the latent space.

[0094] In one or more examples or embodiments, processing the decoder output signal includes using a signal processing encoder to expand the number of channels and reduce the temporal resolution of the decoder output signal, and using a signal processing decoder to reduce the number of channels and expand the temporal resolution of the decoder output signal.

[0095] A method of operating an audio device configured to function as a receiver device, such as an audio device, is disclosed.

[0096] The method includes obtaining an audio input signal from a transmitter device, wherein the audio input signal is an encoded audio input signal encoded based on one or more encoder parameters.

[0097] The method comprises decoding an audio input signal using a decoder and one or more decoder parameters for providing a decoder output signal.

[0098] The method comprises obtaining codec information indicating one or more encoder parameters and / or one or more decoder parameters.

[0099] For example, the method comprises processing a decoder output signal using a first signal processor and based on codec information for providing a first signal processor output signal, wherein, for example, processing the decoder output signal using the first signal processor comprises: processing the decoder output signal using the first signal processor by applying a generative model of processing based on the codec information, for example.

[0100] In one or more example methods, for example, the method includes processing a decoder output signal using an audio modulator of one or more processors of an audio device to provide one or more audio features. In one or more example methods, for example, processing the decoder output signal using a first signal processor includes: processing the decoder output signal based on the one or more audio features.

[0101] In one or more example methods, the one or more audio features include one or more of a Bark parameter and a Mel parameter. In one or more example methods, for example, processing the decoder output signal using the first signal processor includes: processing the decoder output signal based on one or more of the Bark parameter and the Mel parameter.

[0102] In one or more example methods, for example, the method includes determining codec information using a codec adjuster. In one or more example methods, for example, obtaining the codec information includes: determining the codec information using a codec adjuster of one or more processors, and providing the codec information to the first signal processor.

[0103] In one or more example methods, obtaining codec information includes obtaining codec information from a transmitter device.

[0104] In one or more example methods, wherein the codec information includes one or more of a codec type, a sampling rate, and a bit rate. In one or more example methods, for example, processing the decoder output signal using the first signal processor includes: processing the decoder output signal based on one or more of the codec type, the sampling rate, and the bit rate.

[0105] In one or more example methods, wherein the method includes operating a second signal processor in one or more processors according to a non-generative model based on processing of a non-generative signal. In one or more example methods, the method includes processing a second signal processor input signal with the second signal processor to provide a second signal processor output signal. In one or more example methods, the second signal processor input signal is based on a decoder output signal. In one or more example methods, the method includes processing the decoder output signal with the second signal processor.

[0106] In one or more example methods, the method includes determining signal-to-noise ratio information based on the decoder output signal. In one or more example methods, processing the decoder output signal includes processing the decoder output signal based on the signal-to-noise ratio information.

[0107] In one or more example methods, for example, the method includes combining, using a mixer of the audio device, the first signal processor output signal and the second signal processor output signal for providing an audio output signal. In one or more example methods, combining the first signal processor output signal and the second signal processor output signal is based on signal-to-noise ratio information.

[0108] In one or more example methods, the method includes determining a processing scheme for processing a decoder output signal. In one or more example methods, the processing scheme includes a number of iterations of processing and / or noise reduction parameters associated with each iteration of processing. In one or more example methods, for example, processing the decoder output signal using the first signal processor includes: processing the decoder output signal based on the processing scheme.

[0109] In one or more example methods, the first signal processor includes a neural network that is a multi-resolution network. In one or more example methods, the first signal processor includes a signal processing encoder and a signal processing decoder. In one or more example methods, for example, processing the decoder output signal includes using the signal processing encoder to expand the number of channels and reduce the time resolution of the decoder output signal. In one or more example methods, for example, processing the decoder output signal includes using the signal processing decoder to reduce the number of channels and expand the time resolution of the decoder output signal.

[0110] It should be understood that the description of features relating to an audio device configured to act as a receiver also applies to corresponding features in the systems and methods of operating an audio device as disclosed herein.

[0111] A computer-implemented method for training a generative model based on processing of a generative signal is disclosed. In other words, a computer-implemented method for training a generative model based on processing of a generative signal as disclosed herein is disclosed. The method includes obtaining an audio data set including one or more audio signals. The one or more audio signals may be considered as clean audio signals. The audio data set may include multiple audio signals, such as multiple audio signal samples for training.

[0112] The method includes generating a codec-distorted audio data set by encoding one or more audio signals encoded based on one or more encoder parameters (such as using an encoder disclosed herein) to provide one or more encoded audio signals. For example, the method includes generating a codec-distorted audio data set by decoding one or more encoded audio signals using a decoder and one or more decoder parameters to provide one or more codec-distorted audio signals. A codec-distorted audio signal can be considered as an audio signal that has been altered or degraded using a particular codec. For example, a codec-distorted audio signal can be considered as an audio signal that has been compressed or decompressed using a particular codec. The codec-distorted audio data set may include multiple codec-distorted audio signals, such as multiple codec-distorted audio signal samples for training.

[0113] The method includes combining one or more audio signals with one or more white noise signals to provide a white noise audio data set including one or more white noise audio signals. The white noise signal can be considered as a signal including one or more clean audio signals and one or more white noise signals, such as one or more clean audio signals combined with one or more white noise signals. The white noise audio data set may include multiple white noise audio signals, such as multiple white noise audio signal samples for training.

[0114] The method includes determining one or more estimated white noise signals and one or more estimated audio signals by applying a generative model to one or more white noise audio signals and one or more codec distorted audio signals. In other words, the generative model can be configured to identify white noise in the audio signal and generate one or more estimated white noise signals based on the identified white noise. For example, the generative model can be configured to identify one or more audio signals based on the white noise audio signal in one or more white noise signals present in the white noise audio signal. For example, the generative model can be configured to separate one or more audio signals from one or more white noise signals, and then determine one or more estimated white noise signals and one or more estimated audio signals.

[0115] It will be appreciated that the generative model is trained to learn how to discriminate clean audio (eg, one or more audio signals) from white noise (eg, one or more white noise signals) for a variety of white noise amplitudes and / or white noise combinations.

[0116] The method includes training a generative model based on one or more of one or more audio signals, one or more estimated audio signals, one or more estimated white noise signals, and one or more white noise signals.

[0117] In one or more example methods, the method includes obtaining codec information indicating one or more encoder parameters and / or one or more decoder parameters. Then, when used in an operating mode other than a training mode, the obtained codec information can be stored and used by a first signal processor (such as a generative model). It is understood that the codec information can be mapped to different audio data sets, such as a white noise audio data set and / or a codec distortion audio data set. For example, when a generative model is trained on several codecs at the same time, the codec information can be used. It is understood that when a generative model is trained on one type of codec, the codec information can be embedded in encoding and decoding. In one or more examples or embodiments, the generative model can be viewed as a large neural network that can be adjusted based on different codec information, such as trained based on different codec information. Alternatively or additionally, the generative model can be viewed as multiple smaller neural networks, each of which is adjusted based on different codec information, such as each of which is trained based on different codec information.

[0118] It should be understood that the description of features related to an audio device configured to function as a receiver also applies to corresponding features in the system and computer-implemented method for training a generative model based on processing of a generative signal as disclosed herein.

[0119] Figure 1 An example system according to the present disclosure, such as an audio device system 2, is schematically shown. The audio device system 2 can be considered an audio communication system. The system 2 can be considered a communication system for performing a call (such as an audio and / or video call). The system 2 (such as a communication system) can be considered a system that utilizes processing based on codec information. The audio device system 2 can include an audio device 10 (such as an audio device configured to be used as a receiver device as disclosed herein) and an audio device 20 (such as an audio device configured to be used as a transmitter device as disclosed herein).

[0120] The audio device 10 is configured to function as a receiver device. In other words, the audio device is configured to receive an input signal (eg, an audio input signal) from the audio device 20 configured to function as a transmitter device.

[0121] The audio device 10 can be considered as an audio device configured to obtain an audio signal, output an audio signal, and process an audio signal. The audio device 10 can be considered as a conference audio device, for example, the conference audio device is configured to be used by one party (such as one or more users at the near end) to communicate with one or more other parties (such as one or more users at the far end). The audio device 10 can be considered as an intelligent audio device. The audio device 10 can be used for communication, conference, and / or meeting between two or more parties far away from each other. The audio device 10 can be used by one or more users near the location where the audio device 10 is located (also referred to as the near end). In this example, the receiver end can be considered as the near end, and the transmitter end can be considered as the far end.

[0122] The audio device 20 is configured to send the transmitter output signal 78A to an audio device configured to function as a receiver device, such as the audio device 10. The audio device 20 includes an interface, an audio speaker, a microphone 20E, a processor 20C, and a memory. The audio device 10 may be a headset, a computer, a smartphone, etc.

[0123] The audio device 20 is configured to obtain a microphone input signal 70 via a microphone 20E. The microphone input signal 70 may indicate an audio signal generated by a user at a far end (such as a transmitter end). In other words, the microphone input signal 70 may indicate speech, such as speech from a far end transmitter device 20.

[0124] The audio device 20 is configured to determine the transmitter device output signal 78A based on the microphone input signal 70. In one or more embodiments or examples, the microphone input signal 70 may undergo signal processing at the transmitter device 20, such as encoding, compression, and / or enhancement.

[0125] The audio device 20 is configured to output the transmitter output signal 78A via an interface to a receiver device, such as the audio device 10. Outputting the transmitter output signal 78A may include outputting the transmitter output signal 78A to the receiver audio device 10.

[0126] The audio device 10 is configured to obtain an audio input signal from a transmitter device (e.g., an audio device 20), wherein the audio input signal 50A is an encoded audio input signal encoded based on one or more encoder parameters. For example, the audio device 10 is configured to obtain the audio input signal 50A from the transmitter device 20 using a processor 10C and / or via an interface. In one or more example audio devices, the audio device 10 is configured to obtain the audio input signal 50A from a far end (such as a far end party or user). In other words, the transmitter device 20 can be regarded as an audio device at the far end. The audio input signal 50A includes audio. For example, the audio input signal 50A includes one or more audio input signals. In one or more embodiments or examples, the audio input signal 50A is subjected to signal processing at the transmitter device 20, such as encoding, compression, and / or enhancement. The audio input signal 50A can indicate an audio signal generated by a user at the far end. In other words, the audio input signal 50A can indicate speech, such as speech from a far end transmitter device. The audio input signal 50A may be based on and / or considered as a transmitter output signal 78A of the transmitter device 20 .

[0127] The audio device 10 includes a processor 10C. The processor 10C includes a decoder 12A and a first signal processor 18. The decoder 12A can be implemented as hardware and / or software on the audio device 10. The decoder 12A can be regarded as a module configured to convert and / or decode an audio input signal 50A from a transmitter device 20 into a decoder output signal. In other words, the decoder 12A can be configured to decompress the audio input signal 50A from the transmitter device 20. For example, the decoder can be configured to convert and / or decode the input signal 50A from the transmitter device 20 into an audio signal.

[0128] For example, the processor 10C, such as the first signal processor 18 , is configured to operate according to a generative model based on processing of a generative signal.

[0129] The audio device 10 is configured to decode the audio input signal 50A using the decoder 12A and one or more decoder parameters to provide a decoder output signal 52. For example, the audio device 10 is configured to decode the audio input signal 50A by operating the decoder according to one or more decoder parameters. For example, the decoder output signal 52 is based on the audio input signal 50A. In other words, the decoder output signal 52 can be considered as a decoded output signal. The decoder 12A can be configured to process (such as decode) the audio input signal 50A to provide a decoder output signal 52. The audio input signal 50A can also be considered as a decoder input signal.

[0130] It will be appreciated that the decoder 12A may be configured to decode, extract, and / or obtain one or more audio signals and / or first signal parameters from the audio input signal 50A.

[0131] The audio device 10 is configured to obtain codec information 57 indicating one or more encoder parameters and / or one or more decoder parameters. In one or more example audio devices, for example, a decoder output signal 52 decoded using the decoder 12A includes the codec information 57. For example, the audio device 10 may extract the codec information 57 from the decoder output signal 52, for example, using the first signal processor 18 and / or the decoder 12A.

[0132] The audio device is configured to process the decoder output signal 52 using the first signal processor 18 and based on the codec information for providing a first signal processor output signal 56. Processing the decoder output signal 52 using the first signal processor 18 includes processing the decoder output signal 52 using the first signal processor 18 by applying a generative model of processing based on the codec information.

[0133] In one or more example audio devices, the audio device 10 is configured to determine the receiver output 59 based on processing of the decoder output signal 52 .

[0134] In one or more example audio devices, the first signal processor 18 is configured to process a first signal processor input signal 53 for providing a first signal processor output signal 56 .

[0135] For example, the first signal processor input signal 53 is based on and / or corresponds to the decoder output signal 52. In other words, the first signal processor 18 may receive and / or obtain the decoder output signal 52 directly from the decoder 12A.

[0136] In one or more example audio devices, the one or more processors include an audio conditioner 13 configured to process a decoder output signal 52 for providing one or more audio features. The audio conditioner 13 may be configured to provide 51 the one or more audio features to a first signal processor 18. In one or more example audio devices, processing the decoder output signal 52 using the first signal processor 18 includes processing the decoder output signal 52 based on the one or more audio features.

[0137] In one or more example audio devices, the one or more audio features include one or more of Bark parameters and Mel parameters. In one or more example audio devices, processing the decoder output signal 52 using the first signal processor 18 includes processing the decoder output signal 52 based on one or more of the Bark parameters and Mel parameters.

[0138] In one or more example audio devices, the one or more processors include a codec adjuster 15 configured to determine codec information 57. In one or more example audio devices, obtaining the codec information 57 includes determining the codec information 57 using the codec adjuster 15, and providing 54 the codec information to the first signal processor 18.

[0139] In one or more example audio devices, obtaining the codec information 57 includes obtaining the codec information 57 from the transmitter device 20 .

[0140] In one or more example audio devices, the codec information 57 includes one or more of a codec type, a sampling rate, and a bit rate. In one or more example audio devices, processing the decoder output signal 52 using the first signal processor 18 includes processing the decoder output signal 52 based on one or more of the codec type, the sampling rate, and the bit rate.

[0141] In one or more example audio devices, one or more processors include a second signal processor 16, the second signal processor being configured to operate according to a non-generative model based on processing of a non-generative signal, the second signal processor 16 being configured to process a second signal processor input signal 55 for providing a second signal processor output signal 58, wherein the second signal processor input signal 55 is based on a decoder output signal 52, and wherein the audio device 10 is configured to process the decoder output signal 52 using the second signal processor 16.

[0142] The processor 10C includes a second signal processor 16. In one or more example audio devices, the second signal processor 16 is configured to operate according to a non-generative model based on processing of non-generative signals. Thus, both the first signal processor 14 and the second signal processor 16 may be configured to operate according to non-generative and / or generative models having different characteristics that may take into account each other's weaknesses.

[0143] In one or more example audio devices, the second signal processor 16 is configured to process a second signal processor input signal 55 for providing a second signal processor output signal 58. The audio device 10 may be configured to output an audio device output 59 based on the second signal processor output signal 58. In one or more example audio devices, the second signal processor output signal 58 may be considered as the audio device output 59.

[0144] In one or more example audio devices, the second signal processor output signal 58 may be considered as the first signal processor input signal 53. In other words, Figure 1The positions of the first signal processor 18 and the second signal processor 16 can be exchanged.

[0145] In one or more example audio devices, the audio device 10 is configured to determine signal-to-noise ratio information based on the decoder output signal 52, and wherein processing the decoder output signal 52 includes processing the decoder output signal 52 based on the signal-to-noise ratio information.

[0146] In one or more example audio devices, the audio device 10 includes a mixer 17 configured to combine a first signal processor output signal 56 and a second signal processor output signal 58 for providing an audio output signal 59, wherein combining the first signal processor output signal 56 and the second signal processor output signal 58 is based on signal-to-noise ratio information.

[0147] In one or more example audio devices, the audio device is configured to determine a receiver output 59, such as an audio device output signal, based on the first signal processor output signal 56 and / or the second signal processor output signal 58. By combining the processing of the first signal processor 18 and the second signal processor 16, signal processing of the audio input signal 50A, such as the decoder output signal 52, may be improved, such as enhanced.

[0148] In one or more example audio devices, the audio device 10 is configured to determine a processing scheme for processing a decoder output signal 52, wherein the processing scheme includes a number of iterations of processing and / or noise reduction parameters associated with each iteration of processing, and wherein processing the decoder output signal 52 using the first signal processor includes processing the decoder output signal based on the processing scheme.

[0149] In one or more example audio devices, the first signal processor 18 includes a neural network as a multi-resolution network, wherein the first signal processor 18 includes a signal processing encoder 18A and a signal processing decoder 18B, wherein processing the decoder output signal 52 includes using the signal processing encoder 18A to expand the number of channels and reduce the temporal resolution of the decoder output signal 52, and using the signal processing decoder 18B to reduce the number of channels and expand the temporal resolution of the decoder output signal.

[0150] The audio device 10 may be configured to perform FIG. 2A to FIG. 2B Any of the methods disclosed in .

[0151] The processor 10C is optionally configured to execute FIG. 2A to FIG. 2B ,and Figure 3Any of the operations disclosed in (such as any one or more of S106, S108, S110, S112, S114, S116A, S116B, S116C, S118, S120A, S120B, S120C, S120D, S120E, S120F, S120G, S120H, S120I, S120J, S120K, S206A, S206B, S210A).

[0152] The operation of the audio device 10 may be embodied in the form of an executable logic routine (eg, lines of code, software program, etc.) that may be stored on a non-transitory computer-readable medium (eg, memory) and executed by the processor 10C.

[0153] Furthermore, the operation of the audio device 10 may be considered as a method that the audio device 10 is configured to implement. Furthermore, although the functions and operations described may be implemented in software, such functions may also be implemented via dedicated hardware or firmware, or some combination of hardware, firmware and / or software.

[0154] The memory of the audio device may be one or more of a buffer, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory, a random access memory (RAM), or other suitable devices. In a typical arrangement, the memory may include a non-volatile memory for long-term data storage and a volatile memory used as a system memory for the processor 10C. The memory may exchange data with the processor 10C via a data bus. Control lines and an address bus may also exist between the memory and the processor 10C ( Figure 1 Memory is considered a non-transitory computer-readable medium.

[0155] The memory may be configured to store information in a portion of the memory, such as, for example, an input signal, signal parameters, a receiver configuration, a model, a generative model, a non-generative model, codec information, encoder parameters, decoder parameters, audio features, and / or receiver configuration parameters as disclosed herein.

[0156] The audio device 20 may be configured to perform Figure 3 Any of the methods disclosed in .

[0157] The operation of the audio device 20 may be embodied in the form of an executable logic routine (eg, lines of code, software program, etc.) that may be stored on a non-transitory computer-readable medium (eg, memory) and executed by the processor 10C.

[0158] Furthermore, the operation of the audio device 20 may be considered as a method that the audio device 10 is configured to implement, and vice versa. Furthermore, although the functions and operations described may be implemented in software, such functions may also be implemented via dedicated hardware or firmware, or some combination of hardware, firmware and / or software.

[0159] System 2 may be configured to perform FIG. 2A to FIG. 2B ,and Figure 3 Any of the methods disclosed in .

[0160] Furthermore, the operation of system 2 may be considered as a method that system 2 is configured to implement. Furthermore, although the functions and operations described may be implemented in software, such functions may also be implemented via dedicated hardware or firmware, or some combination of hardware, firmware and / or software.

[0161] FIG. 2A to FIG. 2B A flow chart of an example method, such as method 100 , according to the present disclosure is shown.

[0162] A method 100 of operating an audio device configured to function as a receiver device, such as an audio device, is disclosed.

[0163] The method comprises obtaining S102 an audio input signal from a transmitter device, wherein the audio input signal is an encoded audio input signal encoded based on one or more encoder parameters.

[0164] The method comprises decoding S104 an audio input signal using a decoder and one or more decoder parameters for providing a decoder output signal.

[0165] The method comprises obtaining S116 codec information indicating one or more encoder parameters and / or one or more decoder parameters; and

[0166] For example, the method comprises using a first signal processor and processing S120 a decoder output signal based on codec information for providing a first signal processor output signal, wherein, for example, processing S120 the decoder output signal using the first signal processor comprises: processing the decoder output signal using the first signal processor by applying S120A a generative model of processing based on codec information, for example.

[0167] In one or more example methods, for example, the method includes processing S106 the decoder output signal using an audio modulator of one or more processors of the audio device for providing one or more audio features. In one or more example methods, for example, processing S120 the decoder output signal using the first signal processor includes: processing S120B the decoder output signal based on the one or more audio features.

[0168] In one or more example methods, the one or more audio features include one or more of a Bark parameter and a Mel parameter. In one or more example methods, for example, processing S120 the decoder output signal using the first signal processor includes: processing S120C the decoder output signal based on one or more of the Bark parameter and the Mel parameter.

[0169] In one or more example methods, for example, the method includes determining S116A codec information using a codec regulator. In one or more example methods, for example, obtaining the S116 codec information includes determining the S116A codec information using a codec regulator of one or more processors, and providing the S116B codec information to the first signal processor.

[0170] In one or more example methods, obtaining S116 codec information includes obtaining S116C codec information from a transmitter device.

[0171] In one or more example methods, wherein the codec information includes one or more of a codec type, a sampling rate, and a bit rate. In one or more example methods, for example, processing S120 the decoder output signal using the first signal processor includes: processing S120 the decoder output signal based on one or more of the codec type, the sampling rate, and the bit rate.

[0172] In one or more example methods, wherein the method includes operating S108 a second signal processor in one or more processors according to a non-generative model based on processing of a non-generative signal. In one or more example methods, the method includes processing S110 a second signal processor input signal using the second signal processor for providing a second signal processor output signal. In one or more example methods, the second signal processor input signal is based on a decoder output signal. In one or more example methods, the method includes processing the decoder output signal using the second signal processor.

[0173] In one or more example methods, the method comprises determining S112 signal-to-noise ratio information based on the decoder output signal. In one or more example methods, processing S120 the decoder output signal comprises: processing S120E the decoder output signal based on the signal-to-noise ratio information.

[0174] In one or more example methods, for example, the method includes combining, using a mixer of the audio device, the first signal processor output signal and the second signal processor output signal for providing an audio output signal. In one or more example methods, combining the first signal processor output signal and the second signal processor output signal is based on signal-to-noise ratio information.

[0175] In one or more example methods, the method includes determining S118 a processing scheme for processing the decoder output signal. In one or more example methods, the processing scheme includes a number of iterations of the processing and / or noise reduction parameters associated with each iteration of the processing. In one or more example methods, for example, processing S120 the decoder output signal using the first signal processor includes: processing S120F the decoder output signal based on the processing scheme.

[0176] In one or more example methods, the first signal processor includes a neural network as a multi-resolution network. In one or more example methods, the first signal processor includes a signal processing encoder and a signal processing decoder. In one or more example methods, for example, processing the S120 decoder output signal includes using the signal processing encoder to expand the number of S120G channels and reduce the time resolution of the S120H decoder output signal. In one or more example methods, for example, processing the decoder output signal includes using the signal processing decoder to reduce the number of S120I channels and expand the time resolution of the S120J decoder output signal.

[0177] Figure 3 is a flow chart of an example method, such as method 200 , according to the present disclosure.

[0178] A computer-implemented method 200 for training a generative model based on processing of a generative signal is disclosed. The method comprises obtaining S202 an audio dataset comprising one or more audio signals.

[0179] The method comprises generating S206 a codec distorted audio data set by encoding S206A one or more audio signals encoded based on one or more encoder parameters for providing one or more encoded audio signals. For example, the method comprises generating S206 a codec distorted audio data set by decoding S206B one or more encoded audio signals using a decoder and one or more decoder parameters for providing one or more codec distorted audio signals.

[0180] The method comprises combining S208 the one or more audio signals with the one or more white noise signals for providing a white noise audio data set comprising the one or more white noise audio signals;

[0181] The method comprises determining S210 one or more estimated white noise signals and one or more estimated audio signals by applying S210A a generative model to one or more white noise audio signals and one or more codec distorted audio signals; and

[0182] The method comprises training S212 a generative model based on one or more of the one or more audio signals, the one or more estimated audio signals, the one or more estimated white noise signals, and the one or more white noise signals.

[0183] In one or more example methods, the method comprises obtaining S204 codec information indicating one or more encoder parameters and / or one or more decoder parameters.

[0184] Examples of audio devices, systems and methods according to the present disclosure are set out in the following:

[0185] Item 1. An audio device configured to function as a receiver device, the audio device comprising an interface, an audio speaker, and a microphone, the audio device comprising one or more processors and a memory, the one or more processors comprising a decoder and a first signal processor, wherein the first signal processor is configured to operate according to a generative model based on processing of a generative signal, wherein the audio device is configured to:

[0186] obtaining an audio input signal from a transmitter device, wherein the audio input signal is an encoded audio input signal encoded based on one or more encoder parameters;

[0187] decoding an audio input signal using a decoder and one or more decoder parameters for providing a decoder output signal;

[0188] obtaining codec information indicating one or more encoder parameters and / or one or more decoder parameters; and

[0189] The decoder output signal is processed using the first signal processor and based on the codec information to provide a first signal processor output signal, wherein processing the decoder output signal using the first signal processor comprises: processing the decoder output signal using the first signal processor by applying a generative model of processing based on the codec information.

[0190] Item 2. The audio device of item 1, wherein the one or more processors include an audio conditioner configured to process the decoder output signal to provide one or more audio characteristics, and wherein processing the decoder output signal using the first signal processor includes: processing the decoder output signal based on the one or more audio characteristics.

[0191] Item 3. The audio device of any of the preceding items, wherein the one or more processors include a codec regulator configured to determine codec information, and wherein obtaining the codec information includes: determining the codec information using the codec regulator, and providing the codec information to the first signal processor.

[0192] Item 4. The audio device of any one of the preceding items, wherein obtaining codec information comprises obtaining codec information from a transmitter device.

[0193] Item 5. The audio device of any of the preceding items, wherein the codec information includes one or more of a codec type, a sampling rate, and a bit rate, and wherein processing the decoder output signal using the first signal processor includes processing the decoder output signal based on one or more of the codec type, the sampling rate, and the bit rate.

[0194] Item 6. An audio device according to any of the preceding items, wherein the one or more processors include a second signal processor, the second signal processor being configured to operate according to a non-generative model based on processing of a non-generative signal, the second signal processor being configured to process a second signal processor input signal for providing a second signal processor output signal, wherein the second signal processor input signal is based on the decoder output signal, and wherein the audio device is configured to process the decoder output signal using the second signal processor.

[0195] Item 7. The audio device of any preceding item, wherein the audio device is configured to determine signal-to-noise ratio information based on the decoder output signal, and wherein processing the decoder output signal comprises processing the decoder output signal based on the signal-to-noise ratio information.

[0196] Item 8. The audio device of items 6 and 7, wherein the audio device comprises a mixer configured to combine the first signal processor output signal and the second signal processor output signal for providing the audio output signal, wherein combining the first signal processor output signal and the second signal processor output signal is based on signal-to-noise ratio information.

[0197] Item 9. The audio device of any of the preceding items, wherein the audio device is configured to determine a processing scheme for processing a decoder output signal, wherein the processing scheme includes a number of iterations of processing and / or noise reduction parameters associated with each iteration of processing, and wherein processing the decoder output signal using the first signal processor includes: processing the decoder output signal based on the processing scheme.

[0198] Item 10. The audio device of any of the preceding items, wherein the first signal processor comprises a neural network that is a multi-resolution network, wherein the first signal processor comprises a signal processing encoder and a signal processing decoder, wherein processing the decoder output signal comprises using the signal processing encoder to expand the number of channels and reduce the temporal resolution of the decoder output signal, and using the signal processing decoder to reduce the number of channels and expand the temporal resolution of the decoder output signal.

[0199] Item 11. A method (100) of operating an audio device configured to function as a receiver device, the method comprising:

[0200] obtaining ( S102 ) an audio input signal from a transmitter device, wherein the audio input signal is an encoded audio input signal encoded based on one or more encoder parameters;

[0201] Decoding the audio input signal using a decoder and one or more decoder parameters (S104) for providing a decoder output signal;

[0202] obtaining (S116) codec information indicating one or more encoder parameters and / or one or more decoder parameters; and

[0203] The decoder output signal is processed (S120) using the first signal processor and based on the codec information for providing a first signal processor output signal, wherein processing (S120) the decoder output signal using the first signal processor comprises: processing the decoder output signal using the first signal processor by applying (S120A) a generative model for processing based on the codec information.

[0204] Item 12. The method (100) of Item 11, wherein the method comprises processing (S106) the decoder output signal using an audio modifier of one or more processors of the audio device for providing one or more audio features, and wherein processing (S120) the decoder output signal using the first signal processor comprises: processing (S120B) the decoder output signal based on the one or more audio features.

[0205] Item 13. The method (100) of Item 12, wherein the one or more audio features include one or more of Bark parameters and Mel parameters, and wherein processing (S120) the decoder output signal using the first signal processor includes: processing (S120C) the decoder output signal based on one or more of the Bark parameters and the Mel parameters.

[0206] Item 14. The method (100) according to any one of items 11 to 13, wherein obtaining (S116) the codec information comprises determining (S116A) the codec information using a codec adjuster, and providing (S116B) the codec information to the first signal processor.

[0207] Item 15. The method (100) according to any one of items 11 to 14, wherein obtaining (S116) codec information comprises obtaining (S116C) codec information from a sender device.

[0208] Item 16. The method (100) according to any one of items 11 to 15, wherein the codec information comprises one or more of a codec type, a sampling rate, and a bit rate, and wherein processing (S120) the decoder output signal using the first signal processor comprises: processing (S120D) the decoder output signal based on one or more of the codec type, the sampling rate, and the bit rate.

[0209] Item 17. A method (100) according to any one of items 11 to 16, wherein the method comprises operating (S108) a second signal processor of the one or more processors according to a non-generative model based on the processing of the non-generative signal, processing (S110) a second signal processor input signal using the second signal processor for providing a second signal processor output signal, wherein the second signal processor input signal is based on the decoder output signal, and wherein processing (S120) the decoder output signal comprises processing (S120E) the decoder output signal using the second signal processor.

[0210] Item 18. The method (100) according to any one of items 11 to 17, wherein the method comprises determining (S112) signal-to-noise ratio information based on the decoder output signal, and wherein processing (S120) the decoder output signal comprises: processing (S120F) the decoder output signal based on the signal-to-noise ratio information.

[0211] Item 19. The method according to items 17 and 18, wherein the method comprises combining (S122) the first signal processor output signal and the second signal processor output signal using a mixer of the audio device for providing an audio output signal, wherein combining (S122) the first signal processor output signal and the second signal processor output signal is based on signal-to-noise ratio information.

[0212] Item 20. The method according to any one of items 11 to 19, wherein the method comprises determining (S118) a processing scheme for processing a decoder output signal, wherein the processing scheme comprises a number of iterations of processing and / or a noise reduction parameter associated with each iteration of processing, and wherein processing (S120) the decoder output signal using the first signal processor comprises: processing (S120G) the decoder output signal based on the processing scheme.

[0213] Item 21. A method according to any one of items 11 to 20, wherein the first signal processor comprises a neural network as a multi-resolution network, wherein the first signal processor comprises a signal processing encoder and a signal processing decoder, wherein processing (S120) the decoder output signal comprises using the signal processing encoder to expand (S120H) the number of channels and reduce (S120I) the temporal resolution of the decoder output signal, and using the signal processing decoder to reduce (S120J) the number of channels and expand (S120K) the temporal resolution of the decoder output signal.

[0214] Item 22. A computer-implemented method (200) for training a generative model based on generative signal processing, wherein the method comprises:

[0215] Obtaining ( S202 ) an audio data set including one or more audio signals;

[0216] Generating (S206) a codec-distorted audio dataset by: encoding (S206A) one or more audio signals encoded based on one or more encoder parameters for providing one or more encoded audio signals, and decoding (S206B) the one or more encoded audio signals using a decoder and the one or more decoder parameters for providing one or more codec-distorted audio signals;

[0217] combining ( S208 ) the one or more audio signals with the one or more white noise signals for providing a white noise audio data set including the one or more white noise audio signals;

[0218] determining ( S210 ) one or more estimated white noise signals and one or more estimated audio signals by applying ( S210A) a generative model to one or more white noise audio signals and one or more codec-distorted audio signals; and

[0219] A generative model is trained ( S212 ) based on one or more of the one or more audio signals, the one or more estimated audio signals, the one or more estimated white noise signals, and the one or more white noise signals.

[0220] Item 23. The method according to Item 22, wherein the method comprises:

[0221] Codec information indicating one or more encoder parameters and / or one or more decoder parameters is obtained (S204).

[0222] The use of the terms "first", "second", "third" and "fourth", "primary", "secondary", "tertiary", etc. does not imply any particular order, but rather they are included to identify individual elements. Furthermore, the use of the terms "first", "second", "third" and "fourth", "primary", "secondary", "tertiary", etc. does not imply any order or importance, but rather the terms "first", "second", "third" and "fourth", "primary", "secondary", "tertiary", etc. are used to distinguish one element from another. Note that the words "first", "second", "third" and "fourth", "primary", "secondary", "tertiary", etc. are used here and elsewhere for labeling purposes only, and are not intended to represent any specific spatial or temporal ordering. Furthermore, the labeling of a first element does not imply the presence of a second element, and vice versa.

[0223] It is understood that the accompanying drawings include some circuits or operations shown with solid lines and some circuits, components, features, or operations shown with dotted lines. The circuit or operation included in the solid line is the circuit, component, feature, or operation included in the broadest example. The circuit, component, feature, or operation included in the dotted line is an example of another circuit, component, feature, or operation that can be included in or can be a part of the circuit, component, feature, or operation that can be used in addition to the circuit, component, feature, or operation of the solid line example. It should be understood that these operations do not need to be performed in the order presented. In addition, it should be understood that it is not necessary to perform all of the operations. The example operations can be performed in any order and in any combination. It should be understood that these operations do not need to be performed in the order presented. The circuit, component, feature, or operation included in the dotted line can be considered as optional.

[0224] Other operations not described herein may be incorporated into the example operations. For example, one or more additional operations may be performed before, after, simultaneously with, or between any of the described operations.

[0225] Certain features discussed above as separate implementations may also be implemented in combination as a single implementation. Conversely, features described as single implementations may also be implemented separately in multiple implementations or in any suitable subcombination. Furthermore, while features may be described above as functioning in certain combinations, in some cases one or more features from a claimed combination may be deleted from the combination, and the combination may be claimed as any subcombination or variation of any subcombination.

[0226] It should be noted that the word "comprising" does not necessarily exclude the presence of other elements or steps than those listed.

[0227] It should be noted that the word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements.

[0228] It should be noted that the term "indicating ... " may be viewed as "associated with ... ", "related to ... ", "describing ... ", "characterizing ", and / or "defining ". The terms "indicating ... ", "associated with ... ", "related to ... ", "describing ... ", "characterizing ", and "defining " may be used interchangeably. The term "indicating ... " may be viewed as indicating a relationship. For example, weight data indicating weight may include one or more weight parameters.

[0229] It should be noted that the word "based on" may be viewed as "according to" and / or "derived from". The terms "based on" and "according to" may be used interchangeably. For example, a parameter determined "based on" a data set may be viewed as a parameter determined "according to" the data set. In other words, a parameter may be the output of one or more functions that have a data set as input.

[0230] A function may characterize a relationship between input and output, such as a mathematical relationship, a database relationship, a hardware relationship, a logical relationship, and / or other suitable relationship.

[0231] It should also be noted that any reference signs do not limit the scope of the claims, that examples may be implemented at least partially by means of both hardware and software, and that several "means", "units" or "devices" may be represented by the same item of hardware.

[0232] Although features have been shown and described, it should be understood that they are not intended to limit the claimed disclosure, and it will be apparent to those skilled in the art that various changes and modifications may be made without departing from the scope of the claimed disclosure. Therefore, the description and drawings should be regarded as illustrative rather than binding. The claimed disclosure is intended to cover all alternatives, modifications and equivalents.

Claims

1. An audio device configured to function as a receiver device, the audio device comprising an interface, an audio speaker, and a microphone, the audio device comprising one or more processors and a memory, the one or more processors comprising a decoder and a first signal processor, wherein, The first signal processor is configured to operate according to a generative model based on processing of a generative signal, wherein the audio device is configured to: obtaining an audio input signal from a transmitter device, wherein the audio input signal is an encoded audio input signal encoded based on one or more encoder parameters; decoding the audio input signal using the decoder and one or more decoder parameters for providing a decoder output signal; obtaining codec information indicating the one or more encoder parameters and / or the one or more decoder parameters; and Processing the decoder output signal using the first signal processor and based on the codec information to provide a first signal processor output signal, wherein processing the decoder output signal using the first signal processor comprises: processing the decoder output signal using the first signal processor by applying a generative model of processing based on the codec information.

2. The audio device according to claim 1, wherein The one or more processors include an audio conditioner configured to process the decoder output signal to provide one or more audio features, and wherein processing the decoder output signal using the first signal processor includes: processing the decoder output signal based on the one or more audio features.

3. An audio device according to any one of the preceding claims, wherein The one or more processors include a codec adjuster configured to determine the codec information, and wherein obtaining the codec information includes determining the codec information using the codec adjuster and providing the codec information to the first signal processor.

4. An audio device according to any one of the preceding claims, wherein: Obtaining the codec information includes obtaining the codec information from the transmitter device.

5. An audio device according to any one of the preceding claims, wherein The codec information includes one or more of a codec type, a sampling rate, and a bit rate, and wherein processing the decoder output signal using the first signal processor includes: processing the decoder output signal based on one or more of the codec type, the sampling rate, and the bit rate.

6. An audio device according to any one of the preceding claims, wherein The one or more processors include a second signal processor configured to operate according to a non-generative model based on processing of a non-generative signal, the second signal processor configured to process a second signal processor input signal to provide a second signal processor output signal, wherein the second signal processor input signal is based on the decoder output signal, and wherein the audio device is configured to process the decoder output signal using the second signal processor.

7. An audio device according to any one of the preceding claims, wherein: The audio device is configured to determine signal-to-noise ratio information based on the decoder output signal, and wherein processing the decoder output signal comprises processing the decoder output signal based on the signal-to-noise ratio information.

8. The audio device according to any one of claims 6 and 7, wherein: The audio device comprises a mixer configured to combine the first signal processor output signal and the second signal processor output signal for providing an audio output signal, wherein combining the first signal processor output signal and the second signal processor output signal is based on signal-to-noise ratio information.

9. An audio device according to any one of the preceding claims, wherein: The audio device is configured to determine a processing scheme for processing the decoder output signal, wherein the processing scheme includes a number of iterations of processing and / or noise reduction parameters associated with each iteration of processing, and wherein processing the decoder output signal using the first signal processor includes: processing the decoder output signal based on the processing scheme.

10. An audio device according to any preceding claim, wherein: The first signal processor comprises a neural network as a multi-resolution network, wherein the first signal processor comprises a signal processing encoder and a signal processing decoder, wherein processing the decoder output signal comprises: using the signal processing encoder to expand the number of channels and reduce the temporal resolution of the decoder output signal, and using the signal processing decoder to reduce the number of channels and expand the temporal resolution of the decoder output signal.

11. A method (100) of operating an audio device configured to function as a receiver device, the method comprising: Obtaining ( S102 ) an audio input signal from a transmitter device, wherein the audio input signal is an encoded audio input signal encoded based on one or more encoder parameters; Decoding the audio input signal using a decoder and one or more decoder parameters (S104) to provide a decoder output signal; obtaining (S116) codec information indicating the one or more encoder parameters and / or the one or more decoder parameters; and Using a first signal processor and based on the codec information, processing (S120) the decoder output signal for providing a first signal processor output signal, wherein processing (S120) the decoder output signal using the first signal processor comprises: using the first signal processor, processing the decoder output signal by applying (S120A) a generative model of processing based on the codec information.

12. A computer-implemented method (200) for training a generative model based on processing of a generative signal, wherein: The method comprises: Obtaining ( S202 ) an audio data set including one or more audio signals; Generating (S206) a codec-distorted audio data set by: encoding (S206A) the one or more audio signals encoded based on one or more encoder parameters for providing one or more encoded audio signals, and decoding (S206B) the one or more encoded audio signals using a decoder and one or more decoder parameters for providing one or more codec-distorted audio signals; combining ( S208 ) the one or more audio signals with one or more white noise signals for providing a white noise audio data set comprising the one or more white noise audio signals; determining ( S210 ) one or more estimated white noise signals and one or more estimated audio signals by applying ( S210A) the generative model to the one or more white noise audio signals and the one or more codec distortion audio signals; and The generative model is trained ( S212 ) based on one or more of the one or more audio signals, the one or more estimated audio signals, the one or more estimated white noise signals, and the one or more white noise signals.

13. The method according to claim 12, wherein: The method comprises: Codec information indicating the one or more encoder parameters and / or the one or more decoder parameters is obtained (S204).

Images