System and method for anomaly detection in unmarked set of audio recordings

Abstract

The invention relates to a system and method for anomaly detection in an unmarked set of audio recordings. In some implementations, a device may include receiving first and second audio data sets. Further, the device may generate first, second, third, and fourth audio samples. Further, the device may include determining a level of similarity between the first and second audio samples. Further, the device may include combining the first and second audio samples into an audio pair. Further, the apparatus may include training a machine learning model to map the audio samples to hidden space visualizations in view of time and similarity between the first and second audio samples to produce a trained machine learning model. Further, the device may include mapping the third and fourth audio samples in the hidden space visualization by the machine learning model, where placement of the third and fourth audio samples depends on a level of similarity between the third and fourth audio samples.

Description

Technical Field

[0001] The present disclosure relates to image processing utilizing machine learning models to detect anomalies in audio recordings. Background Art

[0002] In certain environments, abnormal sound signals may indicate harmful situations and must be identified quickly to avoid adverse consequences. In factories, acoustic anomalies may indicate a fault in a machine or equipment, which may have expensive or dangerous consequences. In video surveillance applications, acoustic anomalies may represent unusual or unexpected events, such as those related to break-ins, vandalism, and violence. Audio monitoring is inexpensive to deploy and can monitor the condition of an environment or process without direct visual contact. However, in practice, it is difficult and costly to collect large data sets of acoustic anomalies because it may require deliberately destroying several normally functioning machines or components just to collect abnormal data. In the case of video surveillance applications, deliberately generating abnormal events is limited and sometimes not feasible. Summary of the invention

[0003] A system of one or more computers may be configured to perform specific operations or actions by means of installing software, firmware, hardware, or a combination thereof on the system, which in operation causes the system to perform the actions. One or more computer programs may be configured to perform specific operations or actions by means of including instructions that, when executed by a data processing device, cause the device to perform the actions.

[0004] A system of one or more computers may be configured to perform specific operations or actions by means of installing software, firmware, hardware, or a combination thereof on the system, which in operation causes the system to perform the actions. One or more computer programs may be configured to perform specific operations or actions by means of including instructions that, when executed by a data processing device, cause the device to perform the actions.

[0005] In one general aspect, a computer-implemented method may include receiving a first audio data set and a second audio data set from a microphone. The computer-implemented method may also include generating a first audio sample and a second audio sample based on the first audio data set, wherein each of the first and second audio samples is smaller than the first audio data set, and the first audio sample is different from the second audio sample. The method may further include generating a third audio sample and a fourth audio sample based on the second audio data set, wherein each of the third and fourth audio samples is smaller than the second audio data set, and the third audio sample is different from the fourth audio sample. The method may further include determining a similarity level between the first audio sample and the second audio sample. The method may further include: in response to the similarity level between the first audio sample and the second audio sample being higher than a first predetermined threshold, combining the first audio sample and the second audio sample into an audio pair. The method may also include: training a machine learning model based on the audio pair to map the audio samples to a latent space visualization in view of time and the similarity between the first audio sample and the second audio sample, thereby generating a trained machine learning model. The method may further include mapping the third audio sample and the fourth audio sample in the latent space visualization by the trained machine learning model, wherein the placement of the third audio sample and the fourth audio sample depends on the similarity level of the third audio sample and the fourth audio sample determined by the trained machine learning model. Other embodiments of this aspect include corresponding computer systems, apparatuses, and computer programs recorded on one or more computer storage devices, each of which is configured to perform the actions of the method.

[0006] Implementations may include one or more of the following features. A computer-implemented method, wherein mapping of a third audio sample and a fourth audio sample in a latent space visualization creates a cluster, and the method may include: labeling the third audio sample and the fourth audio sample based on the number of clusters in the latent space visualization. The computer-implemented method may include: receiving a fifth audio sample; generating a probability score for the fifth audio sample, wherein the probability score indicates a probability that the fifth audio sample is associated with the cluster; comparing the probability score to a second predetermined threshold; and in response to the probability score being greater than the second predetermined threshold, associating the fifth audio sample with the cluster. The computer-implemented method, wherein given the first audio data set, the first audio sample and the second audio sample do not overlap. The computer-implemented method, wherein training of the machine learning model is performed via a self-supervised contrastive learning objective. The computer-implemented method, wherein the first audio data set and the second audio data set do not include human annotations. The computer-implemented method, wherein mapping of the third audio sample and the fourth audio sample in the latent space visualization creates a cluster, and the method may include: determining a shared attribute between the third audio sample and the fourth audio sample; and labeling the cluster based on the shared attribute of the third audio sample and the fourth audio sample.

[0007] Implementations of the described techniques may include hardware, a method or process, or computer-tangible media. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A system for training a neural network according to the present disclosure is shown.

[0009] Figure 2 A computing device implementing a system for organizing audio data according to the present disclosure is depicted.

[0010] Figure 3 is a flow chart illustrating the process of an example embodiment of an audio anomaly detection framework that can be trained using only unlabeled audio recordings according to the present disclosure.

[0011] Figure 4 is a flow diagram illustrating the process of an example embodiment of an audio anomaly detection framework in operation according to the present disclosure.

[0012] Figure 5 A schematic diagram depicting the interaction between a computer-controlled machine and a control system in accordance with the present disclosure is depicted.

[0013] Figure 6 Depicted is a schematic diagram of a control system configured to control a vehicle, which may be an at least partially autonomous vehicle or an at least partially autonomous robot, in accordance with the present disclosure.

[0014] Figure 7 A schematic diagram depicting a control system according to the present disclosure Figure 8 Depicted is a schematic diagram of a control system configured to detect anomalies in controlling a power tool, such as a drill or driver, in accordance with the present disclosure.

[0015] Fig. 9 Depicted is a schematic diagram of a control system configured to detect anomalies of an automated personal assistant in accordance with the present disclosure.

[0016] Fig.10 Depicted is a schematic diagram of a control system configured to detect anomalies of a monitoring system in accordance with the present disclosure.

[0017] Fig.11 A schematic diagram of a control system configured to detect anomalies of an imaging system (eg, an MRI device, an x-ray imaging device, or an ultrasound device) in accordance with the present disclosure is depicted.

[0018] Fig.12 A flow chart illustrating a process of an example overview of anomaly score calculation based on the present disclosure is depicted. DETAILED DESCRIPTION

[0019] Embodiments of the present disclosure are described herein. However, it is to be understood that the disclosed embodiments are merely examples, and other embodiments may take various and alternative forms. The drawings are not necessarily drawn to scale; some features may be exaggerated or minimized to show the details of a particular component. Therefore, the specific structural and functional details disclosed herein should not be interpreted as limiting, but merely as a representative basis for teaching those skilled in the art to adopt the embodiments in various ways. As will be understood by those of ordinary skill in the art, the various features illustrated and described with reference to any one of the drawings may be combined with the features illustrated in one or more other drawings to produce embodiments that are not explicitly illustrated or described. The combination of the illustrated features provides a representative embodiment of a typical application. However, for a particular application or implementation, various combinations and modifications of features consistent with the teachings of the present disclosure may be desired.

[0020] As used herein, "a," "an," and "the" refer to both singular and plural referents unless the context clearly dictates otherwise. By way of example, a "processor" programmed to perform various functions refers to one processor programmed to perform each function, or more than one processor collectively programmed to perform each of the various functions.

[0021] Contrastive Representation Learning In the present disclosure, a method for discovering latent acoustic states by clustering deep audio representations that can be obtained via self-supervised contrastive learning is described. Specifically, self-supervised contrastive learning can allow for the exposure of manifest acoustic event properties without human interaction. The systems and methods described herein can be configured to embed audio data into a low-dimensional topological space in which recordings characterized by similar properties can be located close to each other and far from other recordings.

[0022] The systems and methods described herein may be configured to utilize nonlinear projection The nonlinear projection can be parameterized as a neural network that Short time segments of the digital audio signal in are mapped to In particular, the systems and methods described herein can train f(·) via back-propagation using a self-supervised contrastive learning objective.

[0023] To learn nonlinear projection The systems and methods described herein can be configured to optimize network parameters, which can minimize the normalized temperature scaled trans-entropy (NT-Xent) loss function.

[0024] The systems and methods described herein may be configured to select N fixed-length context windows, where the systems and methods described herein may assume time invariance for unknown attributes. The systems and methods described herein may randomly sample two segments (e.g., x , x ) from within each context window. i ,x j ), thereby generating a batch of 2N audio clips. The sampling can be performed in the time domain by segmenting the original waveform. In some embodiments, two clips taken from the same context window are considered positive pairs, while the remaining 2(N-1) clips are assumed to be negative examples. The systems and methods described herein can be configured to feed each pair to f(·) in order to extract the corresponding representation. In this way, the systems and methods described herein can optimize the embedding z1...z 2N The following contrastive loss function (or possibly its non-linear projection) is defined above: where {i,j} can be the index of a pair. can be an indicator function that can be zero when l=i and one otherwise. Sim(·,·) can be a similarity measure such as cosine similarity, bilinear similarity, or the inverse of the Minkowski distance, and τ>0 is the softmax temperature that controls the entropy of the logit. The systems and methods described herein can be configured to optimize logτ together with the neural network parameters to avoid manual tuning.

[0025] Density-based clustering The systems and methods described herein may be configured to: after pre-training of f(·) is completed, use the vector-valued representations (z) of audio clips that exhibit similar acoustic features i ,z j ), which can form non-semantic clusters in the latent space without the need for human supervision. These representations can be clustered by one of the following clustering methods: centroid-based methods (such as K-means), distribution-based methods (such as Gaussian mixture models), hierarchical methods (such as link-based algorithms), or any other appropriate clustering methods. However, these clustering concepts typically require ad-hoc interaction with human experts and / or a priori assumptions about the distribution of the target data. Therefore, the systems and methods described herein may prefer density-based methods (e.g., DBSCAN, OPTICS, or any other appropriate density-based clustering method), which may not require a predefined number of clusters K in advance, may involve minimal hyperparameter tuning, and may allow for latent distributions of arbitrary shapes.

[0026] The systems and methods described herein can be configured to treat corresponding cluster indices as annotations of the pseudo-attributes that emerge, and can be used for various downstream tasks, including but not limited to pretext attribute classification. The systems and methods described herein can produce a fully unsupervised pseudo-attribute discovery pipeline that can assign a corresponding cluster index k∈{1,...,K} to each audio clip in a dataset, where K is the total number of clusters identified by the systems and methods described herein.

[0027] Outlier exposure model training The systems and methods described herein can be configured to treat cluster labels as pseudo-attributes in downstream pre-classification tasks. The pre-classifier Θ(·) can be a selected neural network that, given an audio sample at input, predicts the probability that such a sample belongs to each of the K clusters predicted above. Such a neural network can be trained via back-propagation using a cross-entropy loss or any suitable supervised classification loss function.

[0028] The systems and methods described herein may be configured to utilize, at inference time, an audio sample x that may be fed into Θ(·), and the anomaly score for the sample may be calculated as: where p k A(x) is the probability that sample x belongs to the kth cluster, as predicted by Θ(·), where k is the cluster index associated to x by the density-based clustering algorithm as described above. Samples with values ​​of A(x) greater than a specified threshold applied can be identified as anomalies.

[0029] Figure 1 A system 100 for training a neural network is shown. The system 100 may include an input interface for accessing training data 102 for the neural network, where the training data may be raw audio data. Figure 1 As shown, the input interface can be composed of a data storage interface 104, which can access training data 102 from a data storage device 106. In some embodiments, the input interface can be composed of one or more microphones directly or indirectly coupled to the input interface. In some embodiments, the microphone can be a transducer that converts sound into an electrical signal, which can be stored as digital data. For example, the data storage interface 140 can be a memory interface or a persistent storage device interface, such as a hard disk or SSD interface, but can also be a personal area network, local area network or wide area network interface, such as a Bluetooth, Zigbee or Wi-Fi interface or an Ethernet or fiber optic interface. The data storage device 106 can be an internal data storage device of the system 100, such as a hard disk drive or SSD, but can also be an external data storage device, such as a network accessible data storage device; and can be referred to as a non-temporary computer-readable medium.

[0030] In some embodiments, data storage 106 may further include a data representation 108 of an untrained version of the neural network, which may be accessed by system 100 from data storage 106. However, it will be appreciated that training data 102 and data representation 108 of the untrained neural network may each also be accessed from different data storages, such as via different subsystems of data storage interface 104. Each subsystem may be of the type described above with respect to data storage interface 104.

[0031] In some embodiments, the data representation 108 of the untrained neural network may be generated internally by the system 100 based on the design parameters of the neural network, and thus may not be explicitly stored on the data storage device 106. The system 100 may further include a processor subsystem 110 that may be configured to provide an iterated function as a replacement for the layer stack of the neural network to be trained during operation of the system 100. Here, the corresponding layers of the replaced layer stack may have mutually shared weights and may receive the output of the previous layer as input, or, for the first layer of the layer stack, receive an initial activation and a portion of the input of the layer stack.

[0032] The processor subsystem 110 may further be configured to iteratively train the neural network using the training data 102. Here, the iterations of training performed by the processor subsystem 110 may include a forward propagation portion and a backward propagation portion. The processor subsystem 110 may be configured to perform the forward propagation portion by determining an equilibrium point of the iteration function at which the iteration function converges to a fixed point, in addition to other operations that may be performed that define the forward propagation portion—where determining the equilibrium point includes using a numerical root finding algorithm to find a root solution for the iteration function minus its input, and providing the equilibrium point as a surrogate for the output of the stack of layers in the neural network.

[0033] The system 100 may further include an output interface for outputting a data representation 112 of the trained neural network, which data may also be referred to as trained model data. Figure 1 As illustrated, the output interface may be comprised of a data storage interface 104, which in these embodiments is an input / output ("IO") interface, via which the trained model data may be stored in a data storage device 106. For example, a data representation 108 defining an "untrained" neural network may be at least partially replaced by a digital representation 112 of a trained neural network during or after training, as the parameters of the neural network (such as weights, hyperparameters, and other types of neural network parameters) may be adapted to reflect the training on the training data 102. This is also Figure 1 106 by reference numerals 108, 112, which refer to the same data record on the data storage device 106. In some embodiments, the data representation 112 may be stored separately from the data representation 108 defining the "untrained" neural network. In some embodiments, the output interface may be separate from the data storage interface 104, but may generally be of the type described above for the data storage interface 104.

[0034] Figure 2A computing device 200 implementing a system for annotating data is depicted. The computing device 200 may include at least one client device, such as a computing system 202. The computing system 202 may include a processor 204 operably connected to a memory unit 208. The processor 204 may include one or more integrated circuits that implement the functionality of a central processing unit (CPU) 206. The CPU 206 may be a processing unit that implements an instruction set such as one of the x86, ARM, Power, or MIPS instruction set families. During operation, the CPU 206 may execute stored program instructions retrieved from the memory unit 208. The stored program instructions may include software that controls the operation of the CPU 206 to perform the operations described herein. In some examples, the processor 204 may be a system on a chip (SoC) that integrates the functionality of the CPU 206, the memory unit 208, a network interface, and an input / output interface into a single integrated device. The computing system 202 may implement an operating system for managing various aspects of the operation.

[0035] The memory unit 208 may include volatile memory and non-volatile memory for storing instructions and data. The non-volatile memory may include solid-state memory, such as NAND flash memory, magnetic and optical storage media, or any other suitable data storage device that retains data when the computing system 202 is disabled or loses electrical power. The volatile memory may include static and dynamic random access memory (RAM) that stores program instructions and data. For example, the memory unit 208 may store a machine learning algorithm 210 or algorithms, a training data set 212 for the machine learning algorithm 210, and raw source data 215.

[0036] The computing system 202 may include a network interface device 222 configured to provide communications with external systems and devices. For example, the network interface device 222 may include a wired and / or wireless Ethernet interface defined by the Institute of Electrical and Electronics Engineers (IEEE) 802.11 family of standards. The network interface device 222 may include a cellular communication interface for communicating with a cellular network (e.g., 3G, 4G, 5G). The network interface device 222 may further be configured to provide a communication interface to an external network 224 or a cloud.

[0037] The external network 224 may be referred to as the World Wide Web or the Internet. The external network 224 may establish a standard communication protocol between computing devices. The external network 224 may allow information and data to be easily exchanged between computing devices and the network. One or more servers 330 may communicate with the external network 224.

[0038] Computing system 202 may include input / output (I / O) interface 220, which may be configured to provide digital and / or analog input and output. I / O interface 220 may include an additional serial interface (e.g., a universal serial bus (USB) interface) for communicating with external devices.

[0039] The computing system 202 may include a human-machine interface (HMI) device 218, which may include any device that enables the computing device 200 to receive control inputs. Examples of input devices may include human-machine interface inputs, such as keyboards, mice, touch screens, voice input devices, and other similar devices. The computing system 202 may include a display device 232. The computing system 202 may include hardware and software for outputting graphics and text information to the display device 232. The display device 232 may include an electronic display screen, a projector, a printer, or other suitable devices for displaying information to a user or operator. The computing system 202 may further be configured to allow interaction with a remote HMI and a remote display device via a network interface device 222.

[0040] The computing device 200 may be implemented using one or more computing systems. Although this example depicts a single computing system 202 that implements all of the described features, it is intended that the various features and functions may be separated and implemented by multiple computing units that communicate with each other. The specific system architecture selected may depend on various factors.

[0041] The computing device 200 may implement a machine learning algorithm 210 configured to analyze raw source data 215. The raw source data 215 may include raw or unprocessed sensor data that may represent an input data set for a machine learning system. The raw source data 215 may include audio, audio clips, video, video clips, images, text-based information, and raw or partially processed sensor data (e.g., a radar map of an object). In some embodiments, the audio data may be stored in an uncompressed audio format (e.g., WAV, AIFF, AU, PCM, any suitable uncompressed audio format), a lossless compressed audio format (e.g., FLAC, WavPack, ALAC, or any suitable lossless format), and a lossy compressed audio format (e.g., MP3, AAC, Opus, Vorbis, ATRAC, WMA, or any suitable lossy compressed audio format). In some examples, the machine learning algorithm 210 may be a neural network algorithm designed to perform a predetermined function. For example, a neural network algorithm may be configured to determine when an audio signal in an industrial environment is abnormal.

[0042] The computer device 200 may store a training data set 212 for the machine learning algorithm 210. The training data set 212 may represent a previously constructed data set for training the machine learning algorithm 210. The machine learning algorithm 210 may use the training data set 212 to learn weighting factors associated with the neural network algorithm. The training data set 212 may include a source data set, such as audio data, having corresponding achievements or results that the machine learning algorithm 210 attempts to replicate via the learning process. In this example, the training data set 212 may include source audio with and without pedestrians, and corresponding presence and location information. The source video may include various scenes in which pedestrians are identified.

[0043] The machine learning algorithm 210 can operate in a learning mode using the training data set 212 as an input. The machine learning algorithm 210 can be executed in multiple iterations using data from the training data set 212. With each iteration, the machine learning algorithm 210 can update internal weighting factors based on the achieved results. For example, the machine learning algorithm 210 can compare the output results (e.g., annotations) with those output results included in the training data set 212. Since the training data set 212 includes expected results, the machine learning algorithm 210 can determine when the performance is acceptable. After the machine learning algorithm 210 reaches a predetermined performance level (e.g., 100% consistent with the results associated with the training data set 212), the machine learning algorithm 210 can be executed using data that is not in the training data set 212. The trained machine learning algorithm 210 can be applied to a new data set to generate annotated data.

[0044] The machine learning algorithm 210 may be configured to identify specific features in the original source data 215. The original source data 215 may include multiple instances or input data sets for which annotation results are expected. For example, the machine learning algorithm 210 may be configured to identify the presence of pedestrians in an audio waveform and annotate the presence. The machine learning algorithm 210 may be programmed to process the original source data 215 to identify the presence of specific features. The machine learning algorithm 210 may be configured to identify the features in the original source data 215 as predetermined features (e.g., pedestrians). The original source data 215 may be derived from various sources. For example, the original source data 215 may be actual input data collected by the machine learning system. The original source data 215 may be machine generated for testing the system. As an example, the original source data 215 may include raw audio data from a microphone.

[0045] In this example, the machine learning algorithm 210 can process the raw source data 215 and output an indication of a sound representation. The output can also include an enhanced representation of the sound. The machine learning algorithm 210 can generate a confidence level or factor for each generated output. For example, a confidence value exceeding a predetermined high confidence threshold can indicate that the machine learning algorithm 210 is confident that the identified feature corresponds to a particular feature. A confidence value less than a low confidence threshold can indicate that the machine learning algorithm 210 has some uncertainty about the existence of a particular feature.

[0046] Figure 3 is a flow chart illustrating a process 300 of an example embodiment of an audio anomaly detection framework that can be trained using only unlabeled audio recordings according to the present disclosure. According to an example embodiment, one or more process blocks of process 300 can be performed by processor 204 of computing device 200.

[0047] like Figure 3 As shown, process 300 may include receiving a first audio data set and a second audio data set from a microphone (block 302). For example, computing device 200 may receive an audio recording from an industrial environment such as a factory or warehouse from a microphone. In some embodiments, there may be multiple microphones in different locations throughout the industrial environment, thereby creating multiple audio data sets simultaneously. In some embodiments, the microphones are directly linked to computing device 200, or they may be linked to other computing devices, which then send the audio data sets to computing device 200 via network interface device 222. In some embodiments, the audio data sets are uniformly generated to have consistent time intervals (e.g., 10 seconds for each audio data set).

[0048] In addition, Figure 3 As shown, process 300 may include generating a first audio sample and a second audio sample based on a first audio data set, wherein each of the first and second audio samples is smaller than the first audio data set, and the first audio sample is different from the second audio sample (block 304). In some embodiments, computing device 200 may sample the data set and create two or more audio samples from different portions of the data set. For example, a 10-second audio data set may be used to generate two 2-second audio samples, wherein the first two seconds of the audio data set are used to create the first sample, and the last 2 seconds of the audio sample are used to create the second audio sample.

[0049] Likewise Figure 3As shown, process 300 may include generating a third audio sample and a fourth audio sample based on the second audio data set, wherein each of the third and fourth audio samples is smaller than the second audio data set, and the third audio sample is different from the fourth audio sample (block 306). In some embodiments, computing device 200 may sample the data set and create two or more audio samples from different portions of the data set. For example, a 10-second audio data set may be used to generate two 2-second audio samples, wherein the first two seconds of the audio data set are used to create the first sample, and the last 2 seconds of the audio sample are used to create the second audio sample, as described above.

[0050] like Figure 3 As further shown in FIG. 3 , process 300 may include determining a level of similarity between a first audio sample and a second audio sample (block 308). For example, computing device 200 may determine a level of similarity between two audio samples using a similarity metric such as cosine similarity, bilinear similarity, or the inverse of the Minkowski distance. In another embodiment, the similarity may be determined by a pre-trained machine learning model that generates a similarity score between each audio sample. In some embodiments, process 300 may determine that two segments taken from the same audio data set are considered positive pairs, while the remaining 2(N-1) data sets are assumed to be negative examples.

[0051] In addition, Figure 3 As shown, process 300 may include combining the first audio sample and the second audio sample into an audio pair in response to the similarity level between the first audio sample and the second audio sample being above a first predetermined threshold (block 310). In some embodiments, process 300 generates a similarity score between two or more audio samples. The similarity score may be compared to a threshold to determine whether they are similar enough to be considered a positive pair. For example, process 300 may determine that two segments taken from the same audio data set are considered a positive pair, while the remaining data set is assumed to be a negative example, as described above.

[0052] Likewise Figure 3 As shown, process 300 may include training a machine learning model based on the audio pair to map the audio sample to a latent space visualization in view of time and the similarity between the first audio sample and the second audio sample, thereby generating a trained machine learning model (box 312). In particular, the systems and methods described herein may train the machine learning model via backpropagation using a self-supervised contrastive learning objective, as described above. Process 300 may be configured to feed each pair to the machine learning model in order to extract the corresponding representation. In this way, process 300 may train the machine learning model in the embedding z1...z 2N The machine learning model defined above (or possibly its non-linear projection) is: where {i,j} can be the index of a pair. may be an indicator function which may be zero when l=i and one otherwise, as described above.

[0053] like Figure 3 As further shown in , process 300 may include mapping a third audio sample and a fourth audio sample in a latent space visualization by a trained machine learning model, wherein the placement of the third audio sample and the fourth audio sample depends on the similarity level of the third audio sample and the fourth audio sample determined by the trained machine learning model (box 314). In some embodiments, the machine learning model may map the vector value representation of the audio clip in a cluster based on the similarity level of the audio clip, which forms a non-semantic cluster in the latent space without human supervision. These representations may be clustered by one of the following clustering methods: a centroid-based method (such as K-means), a distribution-based method (such as a Gaussian mixture model), a hierarchical method (such as a link-based algorithm), or any other appropriate clustering method. In some embodiments, process 300 may prefer a density-based method (e.g., DBSCAN, OPTICS, or any other appropriate density-based clustering method), which may not require a predefined number of clusters K in advance.

[0054] It should be noted that although Figure 3 An example block diagram of process 300 is shown, but in some implementations, process 300 may include Figure 3 Additional blocks, fewer blocks, different blocks, or differently arranged blocks than those depicted in the process 300. Additionally or alternatively, two or more blocks of the process 300 may be executed in parallel.

[0055] Figure 4 is a flow chart illustrating a process 400 of an example embodiment of an audio anomaly detection framework in operation according to the present disclosure. According to an example embodiment, one or more process blocks of process 400 may be performed by processor 204 of computing device 200.

[0056] like Figure 4 As shown, process 400 may include labeling a third audio sample and a fourth audio sample based on the number of clusters in the latent space visualization (block 402). In some embodiments, the latent space visualization may include multiple clusters, and when an audio sample is placed near an existing cluster, the audio sample will be labeled based on the label of the cluster. In some embodiments, the clusters may be numbered, and the associated labels of the audio samples in the cluster will reflect the number associated with the cluster. In some embodiments, each cluster in the latent visual space may be labeled based on some shared attributes of the audio samples associated with the cluster.

[0057] Likewise Figure 4 As shown, process 400 may include receiving a fifth audio sample (block 404). For example, the device may receive the audio sample from a microphone that records the audio in real time and sends it to computing device 200. In some embodiments, the audio sample may be recorded by the microphone and stored at a remote device, which may transmit the recording to computing device 200 via network interface device 222.

[0058] like Figure 4 As further shown in FIG. 4 , process 400 may include generating a probability score for the fifth audio sample, wherein the probability score indicates a probability that the fifth audio sample is associated with the cluster (block 406). In some embodiments, the second machine learning model generates a probability score that is calculated as follows: where p k A(x) is the probability that sample x belongs to the kth cluster, as predicted by the second machine learning model, where k is the cluster index associated to x by the density-based clustering algorithm as described above. Samples with values ​​of A(x) greater than a second threshold may be identified as anomalies (block 408).

[0059] like Figure 4 As further shown in FIG. 4 , process 400 may include: in response to the probability score being greater than a second predetermined threshold, associating the fifth audio sample with the cluster (block 410). In some embodiments, this may include mapping the vector representation of the fifth audio sample to a latent visualization space that is very close to the cluster. In some embodiments, the fifth audio sample may be labeled based on one or more labels associated with the cluster.

[0060] although Figure 4 An example block diagram of process 400 is shown, but in some implementations, process 400 may include Figure 4 Additional blocks, fewer blocks, different blocks, or differently arranged blocks than those depicted in the process 400. Additionally or alternatively, two or more blocks of the process 400 may be executed in parallel.

[0061] Figure 5A schematic diagram of the interaction between a computer-controlled machine 500 and a control system 502 is depicted. The computer-controlled machine 500 includes an actuator 504 and a sensor 506. The actuator 504 may include one or more actuators, and the sensor 506 may include one or more sensors. The sensor 506 is configured to sense a condition of the computer-controlled machine 500. The sensor 506 may be configured to encode the sensed condition into a sensor signal 508 and transmit the sensor signal 508 to the control system 502. Non-limiting examples of the sensor 506 include video, radar, lidar, ultrasound, and motion sensors. In some embodiments, the sensor 506 is an optical sensor configured to sense an optical image of the environment near the computer-controlled machine 500.

[0062] The control system 502 is configured to receive sensor signals 508 from the computer controlled machine 500. As explained below, the control system 502 may be further configured to calculate actuator control commands 510 depending on the sensor signals and transmit the actuator control commands 510 to the actuators 504 of the computer controlled machine 500.

[0063] like Figure 5 As shown, the control system 502 includes a receiving unit 512. The receiving unit 512 can be configured to receive a sensor signal 508 from the sensor 506 and transform the sensor signal 508 into an input signal x. For example, the sensor 506 can be one or more microphones that can convert sound into a waveform that can be stored as digital data. In an alternative embodiment, the sensor signal 508 is directly received as an input signal x without the receiving unit 512. Each input signal x can be a part of each sensor signal 508. The receiving unit 512 can be configured to process each sensor signal 508 to generate each input signal x. The input signal x can include data corresponding to the sound recorded by the sensor 506.

[0064] The control system 502 includes a classifier 514. The classifier 514 may be configured to classify the input signal x into one or more labels using a machine learning (ML) algorithm, such as the neural network described above. The classifier 514 is configured to be parameterized by parameters such as those described above (e.g., parameter θ). The parameter θ may be stored in and provided by a non-volatile storage device 516. The classifier 514 is configured to determine an output signal y from the input signal x. Each output signal y includes information that assigns one or more labels to each input signal x. The classifier 514 may transmit the output signal y to a conversion unit 518. The conversion unit 518 is configured to convert the output signal y into an actuator control command 510. The control system 502 is configured to transmit the actuator control command 510 to the actuator 504, which is configured to actuate the computer-controlled machine 500 in response to the actuator control command 510. In some embodiments, the actuator 504 is configured to actuate the computer-controlled machine 500 directly based on the output signal y. After the actuator 504 receives the actuator control command 510, the actuator 504 is configured to perform an action corresponding to the associated actuator control command 510. The actuator 504 may include control logic configured to transform the actuator control command 510 into a second actuator control command for controlling the actuator 504. In one or more embodiments, the actuator control command 510 may be used to control a display instead of or in addition to the actuator.

[0065] In some embodiments, the control system 502 includes the sensor 506, which may be an audio sensor such as one or more microphones, instead of or in addition to the computer-controlled machine 500 including the sensor 506. The control system 502 may also include the actuator 504, instead of or in addition to the computer-controlled machine 500 including the actuator 504.

[0066] like Figure 5 As shown, the control system 502 also includes a processor 520 and a memory 522. The processor 520 may include one or more processors. The memory 522 may include one or more memory devices. The classifier 514 (e.g., ML algorithm) of one or more embodiments may be implemented by the control system 502, which includes a non-volatile storage device 516, a processor 520, and a memory 522.

[0067] The non-volatile storage 516 may include one or more persistent data storage devices, such as a hard drive, an optical drive, a tape drive, a non-volatile solid-state device, a cloud storage device, or any other device capable of persistently storing information. The processor 520 may include one or more devices selected from a high-performance computing (HPC) system, including a high-performance core, a microprocessor, a microcontroller, a digital signal processor, a microcomputer, a central processing unit, a field programmable gate array, a programmable logic device, a state machine, a logic circuit, an analog circuit, a digital circuit, or any other device that manipulates signals (analog or digital) based on computer-executable instructions residing in the memory 522. The memory 522 may include a single memory device or multiple memory devices, including but not limited to random access memory (RAM), volatile memory, non-volatile memory, static random access memory (SRAM), dynamic random access memory (DRAM), flash memory, cache memory, or any other device capable of storing information.

[0068] The processor 520 may be configured to read into the memory 522 and execute computer executable instructions residing in the non-volatile storage 516 and embodying one or more ML algorithms and / or methods of one or more embodiments. The non-volatile storage 516 may include one or more operating systems and applications. The non-volatile storage 516 may store computer programs in compiled and / or interpreted form created using various programming languages ​​and / or techniques, including without limitation and alone or in combination Java, C, C++, C#, Objective C, Fortran, Pascal, JavaScript, Python, Perl, and PL / SQL.

[0069] When executed by the processor 520, the computer executable instructions of the non-volatile storage device 516 may cause the control system 502 to implement one or more of the ML algorithms and / or methods disclosed herein. The non-volatile storage device 516 may also include ML data (including data parameters) that support the functions, features, and processes of one or more embodiments described herein.

[0070] The program code embodying the algorithm and / or method described herein can be distributed individually or collectively as a program product in various different forms. The program code can be distributed using a computer-readable storage medium having a computer-readable program instruction thereon, for causing the processor to perform aspects of one or more embodiments. The computer-readable storage medium, which is essentially non-temporary, may include volatile and non-volatile media and removable and non-removable tangible media implemented by any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). The computer-readable storage medium may further include RAM, ROM, erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other solid-state memory technology, portable compact disk read-only memory (CD-ROM) or other optical storage device, cassette, tape, disk storage device or other magnetic storage device, or any other medium that can be used to store desired information and can be read by a computer. The computer-readable program instruction can be downloaded from the computer-readable storage medium to a computer, another type of programmable data processing device or another device, or downloaded to an external computer or external storage device via a network.

[0071] Computer-readable program instructions stored in a computer-readable medium can be used to instruct a computer, other types of programmable data processing devices, or other devices to operate in a particular manner so that the instructions stored in the computer-readable medium produce an article including instructions that implement the functions, actions, and / or operations specified in the flowchart or figure. In certain alternative embodiments, the functions, actions, and / or operations specified in the flowchart and figure can be reordered, processed serially, and / or processed concurrently in accordance with one or more embodiments. Moreover, any of the flowchart and / or figure may include more or fewer nodes or boxes than those illustrated in accordance with one or more embodiments.

[0072] The process, method or algorithm may be embodied in whole or in part using suitable hardware components, such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a state machine, a controller, or other hardware components or devices, or a combination of hardware, software and firmware components.

[0073] Figure 6A schematic diagram of a control system 502 configured to control a vehicle 600, which may be an at least partially autonomous vehicle or an at least partially autonomous robot, is depicted. The vehicle 600 includes an actuator 504 and a sensor 506. The sensor 506 may include one or more microphones. One or more of the one or more specific sensors may be integrated into the vehicle 600. Instead of or in addition to the one or more specific sensors identified above, the sensor 506 may include a software module that is configured to determine the state of the actuator 504 when executed. A non-limiting example of a software module includes a weather information software module that is configured to determine the current or future state of the weather near the vehicle 600 or other location.

[0074] The classifier 514 of the control system 502 of the vehicle 600 may be configured to detect an object near the vehicle 600 depending on the audio signal x. In such an embodiment, the output signal y may include information characterizing that the object is near the vehicle 600. The actuator control command 510 may be determined based on this information. The actuator control command 510 may be used to avoid a collision with the detected object.

[0075] In some embodiments, the vehicle 600 is an at least partially autonomous vehicle, and the actuator 504 may be embodied in the brakes, propulsion system, engine, transmission system, or steering device of the vehicle 600. The actuator control command 510 may be determined so that the actuator 504 is controlled so that the vehicle 600 avoids collision with the detected object. The detected objects may also be classified according to what the classifier 514 thinks they are most likely to be, such as pedestrians or other vehicles. The actuator control command 510 may be determined depending on the classification. In situations where adversarial attacks may occur, the above system may be further trained to better detect objects.

[0076] In some embodiments where the vehicle 600 is an at least partially autonomous robot, the vehicle 600 may be a mobile robot configured to perform one or more functions such as flying, swimming, diving, and walking. The mobile robot may be an at least partially autonomous lawn mower or an at least partially autonomous cleaning robot. In such embodiments, actuator control commands 510 may be determined such that a propulsion unit, a steering unit, and / or a braking unit of the mobile robot may be controlled such that the mobile robot may avoid a collision with the identified object.

[0077] The vehicle 600 may be an at least partially autonomous robot in the form of a household appliance. Non-limiting examples of household appliances include a washing machine, a stove, an oven, a microwave, or a dishwasher. In such a vehicle 600, the sensor 506 may be an audio sensor configured to detect an abnormality associated with the state of an object to be processed by the household appliance. For example, in the case where the household appliance is a washing machine, the sensor 506 may detect an abnormal sound associated with the state of the laundry in the washing machine. The actuator control command 510 may be determined based on the detected abnormal sound.

[0078] Figure 7 A schematic diagram of a control system 502 is depicted, the control system 502 being configured to control a system 700 (e.g., a manufacturing machine), such as a punch cutter, a cutter, or a gun drill, of a manufacturing system 702 (e.g., a portion of a production line). The control system 502 may be configured to control an actuator 504 configured to control the system 700 (e.g., a manufacturing machine).

[0079] The sensor 506 of the control system 700 (e.g., a manufacturing machine) can be a microphone configured to capture one or more auditory properties of the manufactured product 704, such as changes in sound with the operation of the manufacturing machine. The classifier 514 can be configured to determine the state of the manufactured product 704 based on the one or more captured properties. For example, one or more microphones can be used to record audio generated by the manufacturing machine. The recorded audio can be analyzed by a trained machine learning model to detect audio that indicates an abnormal state of the manufacturing machine.

[0080] The actuator 504 may be configured to control the system 700 (e.g., a manufacturing machine) depending on the determined state of the manufactured product 704 for a subsequent manufacturing step of the manufactured product 704. The actuator 504 may be configured to control the function of the control system 700 (e.g., a manufacturing machine) on a subsequent manufactured product 706 of the control system 700 (e.g., a manufacturing machine) depending on the determined state of the manufactured product 704.

[0081] Figure 8 A schematic diagram of a control system 502 configured to control a power tool 800 , such as a drill or driver, having an at least partially autonomous mode is depicted. The control system 502 may be configured to control an actuator 504 configured to control the power tool 800 .

[0082] The sensor 506 controlling the power tool 800 may be a microphone configured to capture one or more auditory properties of the work surface 802 and / or the fastener 804 driven into the work surface 802. In some embodiments, a machine learning model may be trained to determine when the auditory properties received by the one or more microphones indicate an abnormal state of the power tool. The classifier 514 may be configured to determine the state of the work surface 802 and / or the state of the fastener 804 relative to the work surface 802 by one or more of the captured auditory properties. The state may be that the fastener 804 is flush with the work surface 802. Alternatively, the state may be the hardness of the work surface 802. The actuator 504 may be configured to control the power tool 800 so that the driving function of the control power tool 800 depends on the determined state of the fastener 804 relative to the work surface 802 or one or more captured properties of the work surface 802 are adjusted. For example, if the state of the fastener 804 is flush with the work surface 802, the actuator 504 may interrupt the driving function. As another non-limiting example, the actuator 504 may apply additional or less torque depending on the hardness of the working surface 802 .

[0083] Fig. 9 Depicted is a schematic diagram of a control system 502 configured to control an automated personal assistant 900. The control system 502 may be configured to control an actuator 504, which is configured to control the automated personal assistant 900. The automated personal assistant 900 may be configured to control a household appliance, such as a washing machine, stove, oven, microwave, or dishwasher.

[0084] Sensor 506 may be an audio sensor and / or an optical sensor. The audio sensor may be configured to receive a voice command of user 902. The optical sensor may be configured to receive a video image of user 902's gesture 904.

[0085] The control system 502 of the automated personal assistant 900 can be configured to determine an actuator control command 510 that is configured to control the system 502. The control system 502 can be configured to determine the actuator control command 510 based on the sensor signal 508 of the sensor 506, and determine abnormal audio via analysis made by the trained machine learning model. The automated personal assistant 900 is configured to transmit the sensor signal 508 to the control system 502. The classifier 514 of the control system 502 can be configured to execute a gesture recognition algorithm to identify the gesture 904 made by the user 902, determine the actuator control command 510, and transmit the actuator control command 510 to the actuator 504. The classifier 514 can be configured to retrieve information from a non-volatile storage device in response to the gesture 904, and output the retrieved information in a form suitable for receipt by the user 902.

[0086] Fig.10 A schematic diagram of a control system 502 configured to control a monitoring system 1000 is depicted. The monitoring system 1000 can be configured to physically control access through a door 1002. A sensor 506 can be configured to detect scenes relevant to deciding whether to grant access. The sensor 506 can be one or more microphones configured to generate and transmit audio data. The control system 502 can use this data to detect sounds, such as an authorized user entering the area or entering the area at an unauthorized time.

[0087] The classifier 514 of the control system 502 of the monitoring system 1000 can be configured to interpret the audio, image, and / or video data by matching the identities of known persons stored in the non-volatile storage device 516 to determine the identity of the person. The classifier 514 can be configured to generate an actuator control command 510 in response to the interpretation of the audio, image, and / or video data. The control system 502 is configured to transmit the actuator control command 510 to the actuator 504. In this embodiment, the actuator 504 can be configured to lock the door 1002 or unlock the door 1002 in response to the actuator control command 510. In some embodiments, non-physical logical access control is also possible.

[0088] Monitoring system 1000 may also be a surveillance system. In such an embodiment, sensor 506 may be a microphone configured to detect sounds of a scene under surveillance, and control system 502 may be configured to control display 1004. Classifier 514 may be configured to determine the classification of a scene, such as whether an audio signal detected by sensor 506 is suspicious. Control system 502 may be configured to transmit actuator control commands 510 to display 1004 in response to the classification. Display 1004 may be configured to adjust displayed content in response to actuator control commands 510. For example, display 1004 may highlight sounds that classifier 514 considers suspicious. Using embodiments of the disclosed system, a surveillance system may predict the appearance of an object at a certain time in the future.

[0089] Fig.11 A schematic diagram of a control system 502 is depicted, which is configured to control an imaging system 1100, such as an MRI device, an x-ray imaging device, or an ultrasound device. The sensor 506 can be, for example, an audio sensor. The classifier 514 can be a trained machine learning model that is configured to determine abnormal classifications of all or part of the sensed audio recordings. The classifier 514 can be configured to determine or select an actuator control command 510 in response to the classification obtained by the trained neural network. For example, the classifier 514 can interpret a time period of the sensed audio as potentially abnormal. In this case, the actuator control command 510 can be determined or selected so that the display 1102 displays the imaging and highlights the time period of the potential abnormality.

[0090] In some embodiments, a method for marking audio data includes receiving video stream data associated with a data capture environment from at least one camera and microphone. The method also includes receiving audio stream data corresponding to at least a portion of the video stream data from at least one audio capture array. The method also includes marking at least some objects of the video stream data using output from at least a first machine learning model, the first machine learning model being configured to provide output including one or more object detection predictions. The method also includes calculating at least one offset value for at least a portion of the audio stream data corresponding to at least one marked object of the video stream data based on at least one data capture characteristic, and synchronizing at least a portion of the video stream data with the portion of the audio stream data corresponding to at least one marked object of the video stream data using the at least one offset value. The method also includes marking at least the portion of the audio stream data corresponding to at least one marked object of the video stream data using one or more tags of the marked objects of the video stream data and the at least one offset value. The method also includes generating training data using at least some of the marked portions of the audio stream data, and training a second machine learning model using the training data.

[0091] In some embodiments, at least one audio capture array includes a plurality of audio capture devices. In some embodiments, at least one audio capture array is positioned away from at least one image capture device. In some embodiments, using outputs from at least a first machine learning model to mark at least some objects of the video stream data includes marking at least some objects of the video stream data with at least an event type, an event start indicator, and an event end indicator. In some embodiments, at least one data capture characteristic includes one or more characteristics of at least one image capture device. In some embodiments, at least one data capture characteristic includes one or more characteristics of at least one audio capture array. In some embodiments, at least one data capture characteristic includes one or more characteristics corresponding to the position of at least one image capture device relative to at least one audio capture array. In some embodiments, at least one data capture characteristic includes one or more characteristics corresponding to the movement of an object in the video stream data. In some embodiments, based on at least one data capture characteristic, calculating at least one offset value of at least a portion of the audio stream data corresponding to at least one marked object of the video stream data includes using at least one probability-based function.

[0092] Fig.12 A flow chart illustrating a process 1200 of an example overview of anomaly score calculation based on the present disclosure is depicted. The process 1200 may be performed by a processor 204 on a computing device 200. To begin the process 1200, the processor 204 may receive raw audio data (eg, raw source data 216) from one or more microphones.

[0093] At 1202, process 1200 selects N fixed-length context windows, where the systems and methods described herein may assume time invariance for unknown attributes. Process 1200 may randomly sample two audio clips (e.g., x ) from within each context window. i 、x j ), thereby generating a batch of 2N audio clips. The sampling can be performed in the time domain by segmenting the original waveform. In some embodiments, two clips taken from the same context window are considered positive pairs, while the remaining 2(N-1) clips are assumed to be negative examples. Process 1200 can feed each pair to f(·) to extract the corresponding representation. In this way, the systems and methods described herein can optimize the embedding z1...z 2N The following contrastive loss function (or possibly its non-linear projection) is defined above: where {i,j} can be the index of a pair. can be an indicator function that can be zero when l=i and one otherwise. Sim(·,·) can be a similarity measure such as cosine similarity, bilinear similarity, or the inverse of the Minkowski distance, and τ>0 is the softmax temperature that controls the entropy of the logit. Process 1200 can optimize logτ together with the neural network parameters to avoid manual tuning.

[0094] At 1204, process 1200 can utilize vector-valued representations (z) of audio clips exhibiting similar acoustic features after pre-training of f(·) is completed. i ,z j ), which can form non-semantic clusters in the latent space without the need for human supervision. These representations can be clustered by one of the following clustering methods: centroid-based methods (such as K-means), distribution-based methods (such as Gaussian mixture models), hierarchical methods (such as link-based algorithms), or any other appropriate clustering method. However, these clustering concepts typically require ad hoc interaction with human experts and / or a priori assumptions about the distribution of the target data. Therefore, process 1200 can utilize a density-based method (e.g., DBSCAN, OPTICS, or any other appropriate density-based clustering method), which may not require a predefined number of clusters K in advance, may involve minimal hyperparameter tuning, and may allow for latent distributions of arbitrary shapes.

[0095] Process 1200 can treat the corresponding cluster index as an annotation of the pseudo-attribute that emerges and can be used for various downstream tasks, including but not limited to pre-attribute classification. Process 1200 can produce a completely unsupervised pseudo-attribute discovery pipeline that can assign a corresponding cluster index to each audio clip in a dataset, where K is the total number of clusters identified by the systems and methods described herein.

[0096] At 1206, process 1200 can treat cluster labels as pseudo attributes in downstream pre-classification tasks. The pre-classifier Θ(·) can be a selected neural network that, given an audio sample at input, predicts the probability that such sample belongs to each of the K clusters predicted above. Such a neural network can be trained via back-propagation using a cross-entropy loss or any suitable supervised classification loss function.

[0097] At 1208, process 1200 can utilize an audio sample x at inference time that can be fed into Θ(·), and the anomaly score for the sample can be calculated as: where p kA(x) is the probability that sample x belongs to the kth cluster, as predicted by Θ(·), where k is the cluster index associated to x by the density-based clustering algorithm as described above. Samples with values ​​of A(x) greater than a specified threshold applied can be identified as anomalies.

[0098] Although exemplary embodiments are described above, these embodiments are not intended to describe all possible forms covered by the claims. The words used in the specification are descriptive rather than limiting words, and it is understood that various changes can be made without departing from the spirit and scope of the present disclosure. As previously described, the features of the various embodiments can be combined to form further embodiments of the present invention, which may not be explicitly described or illustrated. Although various embodiments can be described as providing advantages or being preferred over other embodiments or prior art implementations with respect to one or more desired characteristics, it is recognized by those of ordinary skill in the art that one or more features or characteristics can be compromised to achieve the desired overall system properties, depending on the specific application and implementation. These properties may include, but are not limited to, cost, strength, durability, life cycle cost, marketability, appearance, packaging, size, serviceability, weight, manufacturability, ease of assembly, etc. Therefore, insofar as any embodiment is described as not as desirable as other embodiments or prior art implementations with respect to one or more characteristics, these embodiments are not outside the scope of the present disclosure and may be desirable for specific applications.

Claims

1. A computer-implemented method for detecting anomalies in audio data, comprising: receiving a first audio data set and a second audio data set from a microphone; generating a first audio sample and a second audio sample based on the first audio data set, wherein each of the first and second audio samples is smaller than the first audio data set and the first audio sample is different from the second audio sample; generating a third audio sample and a fourth audio sample based on the second audio data set, wherein each of the third and fourth audio samples is smaller than the second audio data set and the third audio sample is different from the fourth audio sample; determining a level of similarity between the first audio sample and the second audio sample; In response to a similarity level between the first audio sample and the second audio sample being above a first predetermined threshold, combining the first audio sample and the second audio sample into an audio pair; training a machine learning model based on the audio pairs to map the audio samples to a latent space visualization in view of time and similarity between the first audio sample and the second audio sample, thereby generating a trained machine learning model; as well as The third audio sample and the fourth audio sample are mapped in the latent space visualization by the trained machine learning model, wherein placement of the third audio sample and the fourth audio sample depends on a similarity level of the third audio sample and the fourth audio sample determined by the trained machine learning model.

2. The computer-implemented method of claim 1 , comprising: The third audio sample and the fourth audio sample are labeled based on the number of clusters in the latent space visualization.

3. The computer-implemented method of claim 2, comprising: receiving a fifth audio sample; generating a probability score for a fifth audio sample, wherein the probability score indicates a probability that the fifth audio sample is associated with the cluster; comparing the probability score to a second predetermined threshold; as well as In response to the probability score being greater than a second predetermined threshold, a fifth audio sample is associated with the cluster. 4 . The computer-implemented method of claim 1 , wherein, given the first audio data set, the first audio sample and the second audio sample do not overlap.

5. The computer-implemented method of claim 1, wherein training of the machine learning model is performed via a self-supervised contrastive learning objective. The computer-implemented method of claim 1 , wherein the first audio dataset and the second audio dataset do not include human annotations.

7. The computer-implemented method of claim 1 , wherein mapping of the third audio sample and the fourth audio sample in the latent space visualization creates a cluster, and the method further comprises: determining a shared attribute between the third audio sample and the fourth audio sample; as well as The cluster is labeled based on shared properties of the third audio sample and the fourth audio sample.

8. A system for detecting anomalies in audio data, comprising: One or more processors configured to: receiving a first audio data set and a second audio data set from a microphone; generating a first audio sample and a second audio sample based on the first audio data set, wherein each of the first and second audio samples is smaller than the first audio data set and the first audio sample is different from the second audio sample; generating a third audio sample and a fourth audio sample based on the second audio data set, wherein each of the third and fourth audio samples is smaller than the second audio data set and the third audio sample is different from the fourth audio sample; determining a level of similarity between the first audio sample and the second audio sample; In response to a similarity level between the first audio sample and the second audio sample being above a first predetermined threshold, combining the first audio sample and the second audio sample into an audio pair; training a machine learning model based on the audio pairs to map the audio samples to a latent space visualization in view of time and similarity between the first audio sample and the second audio sample, thereby generating a trained machine learning model; as well as The third audio sample and the fourth audio sample are mapped in the latent space visualization by the trained machine learning model, wherein placement of the third audio sample and the fourth audio sample depends on a similarity level of the third audio sample and the fourth audio sample determined by the trained machine learning model.

9. The system of claim 8, wherein mapping of the third audio sample and the fourth audio sample in the latent space visualization creates a cluster, and the one or more processors are further configured to: The third audio sample and the fourth audio sample are labeled based on the number of clusters in the latent space visualization.

10. The system of claim 9, wherein the one or more processors are further configured to: receiving a fifth audio sample; generating a probability score for a fifth audio sample, wherein the probability score indicates a probability that the fifth audio sample is associated with the cluster; comparing the probability score to a second predetermined threshold; and In response to the probability score being greater than a second predetermined threshold, a fifth audio sample is associated with the cluster.

11. The system of claim 8, wherein with respect to the first audio data set, the first audio sample and the second audio sample do not overlap.

12. The system of claim 8, wherein training of the machine learning model is performed via a self-supervised contrastive learning objective.

13. The system of claim 8, wherein the first audio dataset and the second audio dataset do not include human annotations.

14. The system of claim 8, wherein mapping of the third audio sample and the fourth audio sample in the latent space visualization creates a cluster, and the one or more processors are further configured to: determining a shared property between the third audio sample and the fourth audio sample; and The cluster is labeled based on shared properties of the third audio sample and the fourth audio sample.

15. A non-transitory computer-readable medium storing a set of instructions for detecting anomalies in audio data, the set of instructions comprising: One or more instructions that, when executed by one or more processors of a device, cause the device to: receiving a first audio data set and a second audio data set from a microphone; generating a first audio sample and a second audio sample based on the first audio data set, wherein each of the first and second audio samples is smaller than the first audio data set and the first audio sample is different from the second audio sample; generating a third audio sample and a fourth audio sample based on the second audio data set, wherein each of the third and fourth audio samples is smaller than the second audio data set and the third audio sample is different from the fourth audio sample; determining a level of similarity between the first audio sample and the second audio sample; In response to a similarity level between the first audio sample and the second audio sample being above a first predetermined threshold, combining the first audio sample and the second audio sample into an audio pair; training a machine learning model based on the audio pairs to map the audio samples to a latent space visualization in view of time and similarity between the first audio sample and the second audio sample, thereby generating a trained machine learning model; as well as The third audio sample and the fourth audio sample are mapped in the latent space visualization by the trained machine learning model, wherein placement of the third audio sample and the fourth audio sample depends on a similarity level of the third audio sample and the fourth audio sample determined by the trained machine learning model.

16. The non-transitory computer-readable medium of claim 15, wherein mapping of the third audio sample and the fourth audio sample in the latent space visualization creates a cluster, and the one or more instructions further comprise: The third audio sample and the fourth audio sample are labeled based on the number of clusters in the latent space visualization.

17. The non-transitory computer readable medium of claim 16, the set of instructions further comprising: receiving a fifth audio sample; generating a probability score for a fifth audio sample, wherein the probability score indicates a probability that the fifth audio sample is associated with the cluster; comparing the probability score to a second predetermined threshold; as well as In response to the probability score being greater than a second predetermined threshold, a fifth audio sample is associated with the cluster.

18. The non-transitory computer-readable medium of claim 15, wherein training of the machine learning model is performed via a self-supervised contrastive learning objective.

19. The non-transitory computer-readable medium of claim 15, wherein the first audio data set and the second audio data set do not include human annotations.

20. The non-transitory computer-readable medium of claim 15, wherein mapping of the third audio sample and the fourth audio sample in the latent space visualization creates a cluster, and the one or more instructions further comprise: determining a shared attribute between the third audio sample and the fourth audio sample; as well as The cluster is labeled based on shared properties of the third audio sample and the fourth audio sample.

Images