ANALYSIS OF AN ACOUSTIC SIGNAL

Abstract

A method for analyzing an acoustic signal having a time period and comprising a plurality of repeated audio patterns, comprising the following steps: receiving an audio signal comprising the acoustic signal; determining the repeated audio patterns within the acoustic signal; determining a window length corresponding to a plurality of windows, wherein the window length divides the time period of the acoustic signal into the plurality of windows; and windowing the acoustic signal to obtain the plurality of windows.

Description

ANALYSIS OF AN ACOUSTIC SIGNAL Description The embodiments of the present invention relate to a method for analyzing an acoustic signal and a corresponding apparatus. Other embodiments relate to a system for performing an analysis comprising a respective apparatus. Still other embodiments relate to a computer program. An acoustic signal allows the determination of unwanted effects, such as the deterioration of machinery or an animal disease, such as a non-human mammal, in particular a dog. Las siguientes publicaciones constituyen la técnica anterior: Hebden JH et al., Identification of aortic stenosis and mitral regulation of heartsound analysis, Computers in Cardiology 1997,24:109-112; Zhang Wetal. ,Heartsound classification based on scaled spectrogram and partial least squares regression, Biomedical Signal Processing and Control, 2017,32: 20-28; Ari S et al., Detection of cardiac abnormality from PCG signal using LMS based least square SVM classifier, Expert Systems with Applications, 2010, 37: 8019-8026; Jamous G et al., Optimal time-window duration for computing time / frequency representations of normal phonocardiograms in dogs, Med. & BioL Eng. & Comput., 1992, 30:503-508; e Ismail S et al., Localization and classification of heart beats in phonocardiography signáis - a comprehensive review, EURASIP Journal of Advances in Signal, 2018 (1): 26. In this presentation, it has been found that improvements in acoustic signal analysis can lead to significant improvements in the determination of damage or disease, for example, with regard to accuracy and reliability. Therefore, it is an objective of the present invention to improve acoustic analysis. This objective is achieved through the central theme of independent claims. The embodiments of the present invention provide a method for analyzing an acoustic signal having a time period and comprising a plurality of repeated audio patterns, for example, the periodic sound of a train passing over a railway sleeper or the heartbeat of an animal such as a non-human mammal, in particular a dog. The method comprises the following steps: To receive an audio signal, such as an audio recording, comprising the acoustic signal; Determine the repeated audio patterns within the acoustic signal; IVIA / a / ¿U¿ó / UU I Ί 40 Determine the length of a window corresponding to a plurality of windows, where the window lengths divide the time period of the acoustic signal into the plurality of windows and Window the acoustic signal to obtain a plurality of windows According to one embodiment, the method further comprises the step of analyzing (separately) the respective windows of the plurality of windows. The methods described in this application are based on the finding that an acoustic signal, such as a series of heartbeats or a series of secondary sounds produced by a rotating machine, has a periodicity. By knowing / determining this periodicity, the acoustic signal can be subdivided into a plurality of windows, such that each window comprises at least one of the repeated audio patterns. This allows each of the repeated audio patterns to be analyzed independently of the others, for example, by comparing this audio pattern with a known audio pattern. Alternatively, the repeated audio pattern can be analyzed with respect to the subsequent audio pattern. It should be noted that the repeated audio patterns can be identical, substantially identical, similar, comprise one or more comparable peaks (shape of the respective amplitude plotted over time), and / or comprise one or more comparable peaks (shape of the altitude plotted over time) and comparable amplitude values ​​at the respective time points within the window length, etc. Depending on the implementation, the window length is the same. For example, window lengths can be determined based on the repetition frequency of the repeated pattern. In another approach, the boundary between two patterns is used to determine the window length corresponding to the respective window. This means that each window length corresponding to each window is determined separately. According to one further embodiment, the step of analyzing the respective windows includes performing feature extraction to obtain one or more extracted features that describe the respective pattern (of the window). Depending on the embodiment, the features to be extracted are from the group comprising a named feature, a time-domain feature, and / or a frequency-domain feature. Some examples are: a maximum, a mean, median, standard deviation, variance, skewness, kurtosis, mean absolute deviation, 25th quantile, 75th quantile, entropy, zero crossing rate, crest factor, duration of a first and / or second peak within the pattern, duration between the first and second peaks within the pattern, duration between the second peak of a first pattern and the first peak of a subsequent pattern, mel frequency spectral coefficients, hue chroma, spectral flatness, spectral kurtosis, spectral skewness, spectral slope, spectral entropy, ML / a / ZUZÓ / UU I Ί 40 dominant frequency, bandwidth, spectral centroid, spectral flux, spectral shift, class information, gravity information, position information, race information, weight information, additional information and / or other parameters or another combination thereof. Furthermore, and / or on the other hand, feature extraction may include a step of reducing the range of values ​​corresponding to one or more extracted features so that the range of values ​​corresponding to one or more extracted features is defined between a minimum value (e.g., 0) and a maximum value (e.g., 1). It should be noted that, depending on the implementation, the audio pattern is defined by one or more peaks. Each repeated audio pattern may also be defined by one or more peaks in combination with a baseline where those peaks have an amplitude value at least five times greater than the baseline level. Alternatively, each repeated audio pattern may be defined by a systole and / or diastole, for example, when the acoustic signal is the heartbeat sequence of an animal, such as a non-human mammal, particularly a dog. According to the implementation methods, the method includes the step of normalizing the audio signal. Depending on the implementation, the steps for determining audio patterns, setting a window length, and windowing are either performed automatically or using artificial intelligence. These steps can be implemented, for example, using a decision tree algorithm, a random forest algorithm, a naive Bayes algorithm, an Adaboost algorithm, and / or a support vector machine algorithm. As previously stated, one possible application is the diagnosis of a disease in an animal, such as a non-human mammal, particularly a dog. Therefore, according to the embodiment, the acoustic / audio signal is a recording of a heartbeat sequence from a dog or other non-human animal or mammal, and / or a recording of a heart murmur sequence from a dog or other non-human animal or mammal. Another embodiment features an apparatus for analyzing an acoustic signal having a time period and comprising a plurality of repeating audio patterns. The apparatus comprises an interface for receiving the audio signal comprising the acoustic signal and a processor. The processor is configured to determine the repeating audio pattern within the acoustic signal and to determine a window length corresponding to a plurality of windows, wherein the window lengths divide a time period of the acoustic signal into the plurality of windows. Furthermore, the processor is configured to window the acoustic signal in order to obtain the plurality of windows. Another embodiment features a system for performing an analysis comprising an apparatus and a microphone. According to a preferred embodiment, the system comprises the apparatus and a stethoscope that IVIA / a / ¿U¿ó / UU I Ί 40 comprises a microphone. According to a more preferred embodiment, the system comprises the apparatus and a digital stethoscope comprising a microphone. According to other embodiments, the method described above can be implemented by computer; therefore, one embodiment refers to a computer program. All forms of implementation can be used to medically examine an animal, especially a non-human mammal, such as a dog or a cat, particularly a dog. The embodiments of the present invention are described below with reference to the accompanying figures, in which IVIA / a / ¿U¿ó / UU I Ί 40 Fig. 1a shows a schematic flowchart illustrating a method for analyzing an acoustic signal according to one basic embodiment; Fig. 1b schematically illustrates the input and output signals of the steps described in the context of Fig. 1a according to other embodiments; Fig. 2a and 2b illustrate an example of an acoustic signal to be processed according to one embodiment; Fig. 3 schematically illustrates a certain step in the processing of an acoustic signal to demonstrate the embodiments; Fig. 4a and 4b schematically illustrate problems that occur during the processing of an acoustic signal to explain the embodiments; Fig. 5 illustrates a schematic block diagram of an apparatus for analyzing an acoustic signal; and Fig. 6a-6c illustrate schematic patterns indicating different diseases. The embodiments of the present invention are described below with reference to the accompanying figures, in which identical reference numbers are assigned to objects having identical or similar functions, so that their descriptions are interchangeable and mutually applicable. Figure 1 illustrates Method 100. Method 100 consists of four basic steps and one optional step following the four basic steps. The four basic steps are marked by the reference numbers 110, 120, 130, 140, where the optional step is indicated by the reference number 150. The order shown is the preferred order, although it is not essential. In the first step 110, an audio signal 10 is received (cf. Fig. 1b). The audio signal 10 comprises an acoustic signal 12 having a time period T0 to T6. The acoustic signal 12 comprises a plurality of repeated audio patterns that are indicated by 12a, 12b, and 12c at time points T1, T3, and T5. In the next step, 120, the audio patterns 12a, 12b, and 12c are identified / determined. For example, this determination can be based on an algorithm that finds repetitions within the (audio) signal. This algorithm can be based on artificial intelligence / machine learning algorithms. In the next step, 130, a window length is determined. The window lengths are determined such that they are as long as the single pattern 12a / 12b / 12c. For example, the entire time period T0 to T6 can be divided by the number of determined patterns 12a, 12b, and 12c. By doing so, a window length equal to the window lengths corresponding to each pattern is determined. For example, the window lengths T0 to T2, T2 to T4, and T4 to T6 are determined. Based on this window length, the time period T0 to T6 is subdivided (see step 140 for reference). The result of this windowing step 140 is a plurality of windows indicated by the reference numbers 14a, 14b, and 14c. Here, window 14a comprises pattern 12a, window 14b comprises pattern 12b, and window 14c comprises pattern 12c. After that, you can proceed to the optional step of analyzing 150. Here, windows 14a, 14b, and 14c are analyzed. For example, window 14b is extracted and analyzed independently of the other windows, for example, by performing feature extraction. This feature extraction can also be performed on windows 14a and 14c. Furthermore, window 14b can be compared with the other windows, for example, windows 14a and 14b, to determine the regularity of the patterns. With regard to Figure 2a, an analysis of a heartbeat sequence is described as an example. Figure 2a shows an audio signal 10' comprising an acoustic signal 12'. This acoustic signal represents, for example, a heartbeat sequence, such as that of a dog. The duration of the recording 10' can be approximately 10 seconds, where these 10 seconds can comprise 11 heartbeat patterns indicated by the reference numbers 12a', 12b', etc., to 12k'. Alternatively, the duration can be at least 3 seconds, at least 5 seconds, at least 15 seconds, at least 30 seconds, or at least 1 minute; generally from 5 to 180 seconds or from 1 to 300 seconds or at least 1 or at least 10 seconds. Each pattern 12a', 12b1 may comprise two signals S1 and S2, as illustrated in figure 2b. Figure 2b illustrates a magnified view of, for example, the two patterns 12a' and 12b'. The S1 signal can be the peak, indicative of the beginning of systole, while S2 is a peak indicative of the beginning of diastole. Each S1 and S2 peak has a comparatively high amplitude compared to the baseline signal. As can be seen, the fundamental structure of pattern 12a' is comparable to the structure of pattern 12b'. This means that the amplitude of the S1 peak is approximately the same when the time interval between S1 and S2 is also comparable in the two patterns 12a' and 12b'. As can be seen with respect to Figure 2b, each 12a' and 12b' pattern can have two or more S1 and S2 peaks. Furthermore, the pattern can have a baseline / zero signal between the two S1 and S2 peaks. The combination of the two peaks S1 and S2 and the base signal between the two peaks and / or the base signal after the last peak S2 can define the pattern.At this stage, it is worth mentioning that a 12a', 12b', etc. pattern can also be defined by a single peak and a base signal, or just two peaks with no base signal in between, or by another combination. In order to separate the patterns 12a', 12b', etc., windowing is performed. From this, the window lengths are determined. The window lengths can be determined based on the duration of the acoustic signal 12', in this case 10 seconds, and the number of patterns 12a', etc., in this case 11 patterns. The calculation can be performed by simple division. In this example, the result would be that the window length corresponding to each window amounts to approximately 0.9 seconds. Of course, the window lengths can be determined differently, according to other embodiments, for example, by determining the duration of each pattern, i.e., the interval between S1 and the subsequent S1, and averaging these durations. According to other embodiments, the window lengths can vary over time, for example, when the periodicity of the pattern varies.This can happen, for example, when the heart rate decreases in the current situation. In this example, the window lengths WL corresponding to all patterns 12a' to 12k' are equal. Therefore, windows 14a' to 14k' are used to subdivide the audio signal 12'. In this way, each window 14a' to 14k' comprises a respective pattern 12a' to 12k'. This makes it easier to extract features within each window 14a' to 14k'; that is, not for the entire recording 10' of the acoustic signal 12', but for each pattern 12a' to 12k', or each heartbeat, respectively. According to the embodiments, the window length can be adapted, e.g., from a first window to a second window (a subsequent window in the plurality of windows). It is resolved that each window length, or the window length of at least two windows, is different / varied. According to the embodiments, an adaptation can be made based on the determination of a heartbeat sound, such as a systole (S1) or another characteristic feature within the pattern or current heart rate of the animal or non-human mammal, such as a dog or cat. Therefore, the method may optionally comprise a step of determining a characteristic feature. ML / a / ZUZÓ / UU I Ί 40 of the pattern (heartbeat) or the heartbeat frequency in order to adapt the window length. Consequently, the window length depends on the heartbeat frequency. One result may be that the lengths of the heartbeat phase / pattern can be determined. Depending on the method used, this may be to obtain comparable windows within which the analysis can be performed, thus determining the respective position of systole (S1, S2) or diastole within the corresponding window to improve the analysis. The position of the murmur within the window / pattern / phase of the heartbeat is a relevant factor. According to some implementations, dynamic windowing can be performed using a wavelet transform. For example, the S1 and S2 peaks within the audio signal are precisely determined so that each window can be set at a specific position relative to that S1 or S2 peak, for example, at the beginning of each peak (on a rising slope). In this way, according to some implementations, the start of each window is determined based on that peak or on a characteristic element of the pattern. According to other implementations, the respective end of each window is determined analogously, for example, at the beginning of the next comparable feature, such as the S1 system.This means that, according to the implementation methods, windowing is carried out by determining a respective feature within each pattern, where the feature of the first pattern is used as the beginning of a respective window, while the end of that window is determined on the basis of the respective feature of the subsequent window. Depending on the specific procedure, the position of the murmur within the diagnostic window provides additional information about the condition. Therefore, the method also includes determining the position (position in time) of the murmur within the respective diagnostic window. For example, it can be differentiated whether the murmur is detected between the first systole (S1) and the second systole (S2), closer to the first systole (S1) and then to the second systole (S2), closer to the second systole (S2) and then to the first systole (S1), or after the second systole (S2).Examples of this type of diagnosis include: a systolic murmur produced by the mitral valve in the left heart chamber (mitral valve regurgitation / leakage), a systolic murmur due to a leaking tricuspid valve (right heart chamber), a systolic murmur resulting from aortic or pulmonic artery stenosis, or a persistent murmur (both systolic and diastolic) resulting from congenital conditions such as patent ductus arteriosus, chamber defects, or atrial wall defects. Murmurs can also be auscultated during diastole, such as those caused by valvular stenosis and / or valvular insufficiency. These different diagnoses can be easily distinguished by their characteristic sound patterns. ML / a / ZUZÓ / UU I Ί 40 Figures 6a-6c illustrate examples of patterns that indicate different diseases. Figure 6a shows a typical pattern corresponding to mitral regurgitation (MR). In this case, the murmur is between S1 and S2. Figure 6b illustrates a murmur due to pulmonic stenosis (PS). In this case, the sound has a crescendo-decrescendo character and extends through 50-85% of systole. Fig. 6c illustrates a patent ductus arteriosus with left-to-right blood shunting (PDA): During systole, reduced, diminished, or even reversed shunting may occur. Another murmur is called dilated cardiomyopathy (DCM): Some dogs with DCM do not produce a sound, but an extension of the ring around the atrial valve can cause mitral and tricuspid regurgitation that has a systolic sound (peak intensity over the apex of the heart). These are typical murmurs that can be determined using the algorithm. According to other implementations, different machine learning strategies can be used to categorize patterns. Some examples include random forests, support vector machines, neural networks, decision trees, and AdaBoost. For instance, detecting mitral valve disease is preferably done using a random forest or AdaBoost algorithm, whereas for other diseases, the algorithm used may vary. According to the embodiment, this additional information is determined automatically. Therefore, the method comprises the step of determining a diagnosis of the respective murmur / disease based on the position of the murmur within the acoustic window or, more generally, based on the structure of the acoustic pattern. According to the embodiment, the pattern is determined within a sequence of repeated patterns or patterns extracted / separated from the sequence, where the sequence comprises a plurality of patterns that are identical or comparable to each other. According to other implementations, feature extraction can be performed on each window from 14a' to 14k' as described below. For example, a range value can be extracted as a feature. Afterward, the feature can be processed, for example, by calculating the average / median value. In short: Determining the edges of the window in relation to the respective record Determination of one or more characteristics per window IVIA / a / ¿U¿ó / UU I Ί 40 Processing one or more features, for example, determining the average, determining the median Continue to the next record Next, with respect to Figure 3, a feature extraction corresponding to the maximum of the feature is described. Figure 3 illustrates an audio recording 10 comprising an acoustic signal 12 that is subdivided into a plurality of windows 14a to 14o. For simplification purposes, only windows 14h and 14m, indicated by the grid 14x, are considered. In this case, the maximum peak amplitudes are determined. These maximum amplitudes are indicated by the reference numbers 16m1 to 16m5. The maximum 16m4” reaches approximately 100 and belongs to window 141”. Since this maximum 14m is significantly higher compared to the maximums 14m1, 14m2, 14m3, and 14m5, and since the entire signal within window 141 appears to be altered, these values ​​are disregarded. The maximum feature of windows 14h, 14i, 14j, and 14m is approximately (15,000, 17,000, 19,000, 10,000). The mean feat_max is 15,215, while the median feat_max is 16,000.Instead of reducing the four characteristics 16m1, 14m2, 14m3 and 14m5, all four characteristics can also be used. As illustrated with respect to Figure 3, some patterns within the window are not usable. This is illustrated with respect to Fig. 4a. Fig. 4a shows an audio signal 10' comprising an acoustic signal 12' that must be subdivided into a plurality of windows. Here, some windows 14a' to 14o' are illustrated. In particular, in the case of windows 14a', 14h', and 14o', the respective pattern 12a', 12h', and 12o' is disturbed. Therefore, these windows 14a', 14h', and 14o' are not used / are omitted. Unlike the disturbed window 141 in Fig. 3, which has a maximum value that is too high, the reason for omitting windows 14a', 14h', and 14o' is that the boundary between the preceding and subsequent windows cannot be determined. Since such disturbed signals from windows 14a', 14h', and 14o' or 141 would distort the result of the feature extraction, the respective windows 14a', 14h', 14o', and 141' are omitted. Regarding the embodiment of the Figure.4a, this means that all windows 14a', 14h' and 14o' have been omitted, for which a clear boundary cannot be determined. Figure 4b illustrates another disturbance. In this case, the entire acoustic signal 12 comprises amplitude values ​​that are out of range and have no clear signals, so windowing cannot be performed. Therefore, in some cases, the entire recording 10 comprising the completely disturbed signal 12 without windowing can be omitted. IVIA / a / ZUZÓ / UU I 140 The following describes a possible analysis step for the respective windows 14a, 14b, and 14c (refer to Fig. 1) or for other windows belonging to other embodiments. Windowing allows for the separate analysis of each window 14a, 14b, and 14c and each pattern 12a, 12b, and 12c. Depending on the embodiment, three different types of features can be extracted: name features, time-domain features, and frequency-domain features. The time-domain and frequency-domain features primarily relate to the acoustic signal 12a, 12b, and 12c, while the name feature relates to peripheral information. The possible time-domain features that can be analyzed for each window 14a, 14b, and 14c are: The pattern mean, pattern median, standard deviation within the pattern, variance within the pattern or with respect to another pattern, pattern skewness, pattern kurtosis, mean absolute deviation of the pattern, 25th quantile of the pattern, 75th quantile of the pattern, pattern entropy, zero crossing speed within the pattern, search factor, duration of the first peak as 1, duration of the other peak S2, duration from the end of S1 to the beginning of S1, duration from the end of S2 to the beginning of S1 to the beginning of the next S1. Especially, the duration characteristics are most significant when the signal 12 is windowed into windows 14a to 14c. Time domain characteristics can include cepstral coefficients at the mel frequency, tonal chroma, spectral flatness, spectral kurtosis, spectral slope, spectral entropy, dominant frequency, bandwidth, spectral centroid, spectral flux, and / or spectral shift. As noted earlier, an example of an audio signal could be the heartbeat of an animal or a non-human mammal, such as a dog. Because of this, it is possible to include additional information, namely, so-called naming characteristics. Naming characteristics can include class, severity (disease severity on a scale of 0-6), position (measurement position on the animal, e.g., anterior left, anterior right, posterior left, posterior right), breed, and weight (weight classes can be used, e.g., 0-10 kg, 10-20 kg, >20 kg). Additionally, a node can be included (e.g., post-operative or pre-operative). It should be noted that the lists of different types of features and feature types are not limited to those mentioned. Depending on the implementation, the described analysis step is performed primarily or entirely automatically. Specifically, windowing can be executed automatically. During the learning phase, windowing and a sample analysis can be performed. In the learning phase, parameters for windowing / automatic windowing can be set, including the feature list (specifically for windowing purposes; in this case, the mean / median or the total can be used for the analysis, and the trial size can be the percentage of the trial dataset (e.g., 0.3)). Here, the trial dataset is divided. IVIA / a / ¿U¿Ó / UU I Ί 40 and randomizes, and for example, 30% is used for learning. In the case of learning, different models can be used, for example, a decision tree model, a random forest model, a naive Bayes basis model, an AdaBoost model, and / or a support vector machine model. The random forest model has been found to yield the best results. As noted earlier, the described approach can be used for the diagnosis of an animal or non-human mammal, such as a dog. Background information follows. Mitral valve endocardiosis is the most common heart disease in dogs. Prevalence increases with age: approximately 10% of all dogs aged 5 to 8 years, approximately 25% of all dogs aged 9 to 12 years, and 35% of all dogs over 13 years of age are affected. Older dogs of small breeds, such as toy poodles, miniature schnauzers, Yorkshire terriers, and dachshunds, are primarily affected. Another predisposed breed is the Cavalier King Charles Spaniel. This breed is particularly susceptible to mitral valve endocardiosis at a young age. Larger dogs are affected much less frequently. Signs of early-stage disease: Heart murmur: This heart murmur is audible to the veterinarian with a stethoscope, even before the owner notices any changes in their pet. This is why this condition can sometimes be detected during routine examinations, such as vaccination checkups. Signs of illness in the later stages: Cough, increased respiratory rate, difficulty breathing, lethargy, poor performance, loss of appetite, brief periods of unconsciousness. Causes: due to a very irregular heartbeat, a severe cough, or as a result of a tear in the left atrium. Based on the above technique, three diagnostic solutions are known: • X-ray or Heart Size: Initially there is an enlargement of the heart's shadow in the area of ​​the left atrium and later also in the area of ​​the left ventricle. or Dislocation of the left primary bronchus. Another important task of X-ray imaging is the evaluation of the pulmonary vessels and lung fields. If the pulmonary veins are congested, this is an indication for therapy. If pulmonary edema is present, alveolar opacity can be demonstrated, usually in the hilar region. or Pulmonary congestion: At first the pulmonary veins appear congested, later pulmonary edema (water in the lungs) can be diagnosed. • Electrocardiogram or ECG. The ECG primarily diagnoses cardiac arrhythmias. It is an important diagnostic criterion, as dogs with mitral valve disease may have arrhythmias. The usefulness of an ECG is, IVIA / a / ¿U¿ó / UU I Ί 40 Ultimately, it is a decision of the cardiologist, but one should always be performed if an arrhythmia or additional heart sounds are detected during monitoring. • Echocardiography or The size of the atrium and ventricle can be measured to reliably determine any possible enlargement. Or, the ability of the heart muscle to contract can be measured. Additionally, color Doppler echocardiography can be used to quantify the degree of insufficiency. According to the embodiments of the present invention, it is possible to automatically differentiate between pathological and healthy heart murmurs in non-human animals or mammals, particularly dogs. Heart sounds were auscultated in each dog in four different positions (left anterior, left posterior, right anterior, right posterior). From the sound recordings, various features were calculated in both the time and frequency domains, which served as input for several machine learning algorithms after dividing the total dataset into training and test datasets. Unfortunately, the classification between pathological and healthy heart sounds did not yield satisfactory results. For this reason, the classification was initially limited to two classes. These consisted of recordings from dogs with healthy hearts and from dogs with mitral valve regurgitation (MR).Recurrent laryngeal malformation (RLM) is a heart valve defect that causes blood to flow backward from the left ventricle into the left atrium. Using the Decision Tree algorithm, the classification of dogs weighing less than 20 kg achieved 84% accuracy, 81% precision, and 81% repeatability (initial test results). Accuracy is expected to increase. By using additional sound samples, an increase to 93% has been achieved. It should be noted that the dataset is very small. Some breeds are represented only by recordings of dogs with RLM. Using the breed characteristic would lead to skewed results and was therefore not used.Therefore, these methods allow, for example, the diagnosis of heart disease in animals or non-human mammals, particularly dogs, using simple equipment (e.g., a digital stethoscope and smartphone), which is inexpensive and quick to perform (with minimal stress for the animal). This also advantageously enables remote medical examinations. Using the approach described above, a simple apparatus can be formed. The apparatus is illustrated in Figure 5. Figure 5 shows the apparatus 30 comprising at least one processor 32 and one microphone 34. The microphone signal can be digitized using an ADC. In a simple configuration, the apparatus can be formed without a microphone and instead has an interface for receiving the audio signal. This configuration is not illustrated. The processor 32 is configured to determine the repeated audio pattern within the acoustic signal and to determine a window length corresponding to a plurality of windows. The window lengths divide a time period of an acoustic signal into the plurality of windows (uniformly or irregularly). The processor is further configured to window the acoustic signal (from the window lengths) and to obtain the plurality of windows. Moreover, the The IVIA / a / ¿U¿ó / UU I Ί 40 processor can be configured to perform the analysis, for example, by extracting pattern features within a plurality of windows. The entire device can be implemented as a smart device / smartphone, and the analysis can be performed uniformly. A more sophisticated implementation could be a device such as a stethoscope or a digital stethoscope. According to the embodiments, the apparatus described above can be implemented by an intelligent device, such as a smartphone, tablet, personal computer, or other device comprising a processor. The method, or at least some steps of the method defined above or defined in the context of the following embodiments, can be executed using the processor. The method can be implemented, for example, in the form of software, an application, or an algorithm for the intelligent device. Depending on the specific device, the stethoscope can generate a report, such as a diagnostic report. For example, the report might include a diagnosis describing a particular disease or murmur. The report can be summarized using a three-color traffic light system: yellow, red, and green. Green might indicate that no disease or murmur was detected, so the animal / non-human mammal / dog is in good condition. Yellow might indicate a risk or high probability of a murmur or disease. Yellow might also indicate that further monitoring or testing of the animal / non-human mammal / dog is required. Red might indicate that a murmur or disease was detected, so treatment is required or recommended for the animal / non-human mammal / dog. According to a different embodiment, the report may consist of the following: Green light: It is very unlikely that the animal / non-human mammal / dog has a heart murmur. The heart sound is more likely physiological. Yellow light: The heart sound differs slightly from a physiological sound. The animal / non-human mammal / dog may have a heart and / or vascular condition other than a leak / mitral valve disease. Please also consider repeating the auscultation. Red light: The animal / non-human mammal / dog is very likely to have a mitral valve murmur typical of mitral valve disease. The murmur is loud and considered clinically relevant: further testing is recommended, such as a cardiac check-up including a chest X-ray or echocardiogram to determine if the heart is enlarged and requires therapy (stage B2). Option B: The heart sound is soft: a cardiac check-up every 6 months or annually is recommended. IVIA / a / ¿U¿ó / UU I Ί 40 Depending on the implementation methods, a simple summary can be issued. An example of such a summary might be the following: This is not a medical diagnosis. A visit to the veterinarian is recommended for a medical diagnosis. The detection of heart murmurs revealed the following: - with a 95% probability it is a mitral regurgitation (MMVD). - with a 78% probability is Pulmonary Stenosis (PS) II Depending on the implementation method, another type of report is also possible. It should be noted that this report / diagnosis is generated automatically. The following describes other forms of realization in the context of the clauses. Clause 1: A method (100) for analyzing (150) an acoustic signal (12a, 12b and 12c) having a time period (T0 to T6) and comprising a plurality of repeated audio patterns, comprising the following steps: receive (110) an audio signal (10,10', 10, 10', 10) comprising the acoustic signal (12a, 12b and 12c); determine (120) the repeated audio patterns within the acoustic signal (12a, 12b and 12c); determine (120) a window length corresponding to a plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o), where the window length divides the time period (T0 to T6) of the acoustic signal (12a, 12b and 12c) into the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o) and window (140) the acoustic signal to obtain the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o). Clause 2: The method (100) in accordance with clause 1, wherein the method (100) comprises the additional step of analyzing (150) the respective windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o). Clause 3: The method (100) in accordance with clause 2, wherein the additional analysis step (150) comprises the step of performing a feature extraction to obtain one or more extracted features that describe the respective pattern. IVIA / a / ZUZÓ / UU I 140 Clause 4: The method (100) in accordance with clause 3, wherein the features to be extracted are from the group comprising the name feature, the time-domain feature, and / or the frequency-domain feature;and / or where the feature to be extracted is from the group comprising a maximum, a mean, median, standard deviation, variance, skewness, kurtosis, mean absolute deviation, 25th quantile, 75th quantile, entropy, zero crossing speed, crest factor, duration of a first peak and / or second peak within the pattern, duration between the first peak and the second peak within the pattern, duration between the second peak of a first pattern and the first peak of a subsequent pattern, cepstral coefficients at mel frequency, hue chroma, spectral flatness, spectral kurtosis, spectral skewness, spectral slope, spectral entropy, dominant frequency, bandwidth, spectral centroid, spectral flux, spectral shift, class information, severity information, position information, breed information, weight information, additional information and / or other parameters or a combination thereof;and / or in which the feature extraction step comprises the step of redefining the range of values ​​corresponding to one or more extracted features so that the range of values ​​corresponding to one or more extracted features is defined between a minimum value or 0 and a maximum value or 1.; Clause 5: The method (100) in accordance with any of the preceding clauses, wherein the repeated audio patterns are equal to each other, substantially equal to each other, similar to each other, comprise one or more peaks of a comparable shape of the respective amplitude plotted over time and / or comprise one or more peaks of a comparable shape of the amplitude plotted over time and comparable amplitude values ​​at the respective time point within the length of the window. Clause 6: The method (100) in accordance with any of the preceding clauses, wherein the window length is equal. Clause 7: The method (100) in accordance with any of the preceding clauses, wherein the length of the window is determined based on the frequency of repetition of the repeated pattern. Clause 8: The method (100) in accordance with any of the preceding clauses, wherein method (100) further comprises the step of ignoring one or more windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o) without an audio pattern similar or equal to the plurality of repeated audio patterns. Clause 9: The method (100) in accordance with any of the preceding clauses, wherein each repeated audio pattern is defined by one or more peaks and / or IVIA / a / ¿U¿ó / UU I Ί 40 in which each repeated audio pattern is defined by one or more peaks in combination with a base level, in which one or more peaks have an amplitude value that is at least five times greater than the base level and / or where each repeated audio pattern is defined by a systole and / or diastole. Clause 10: The method (100) in accordance with any of the preceding clauses, wherein the method (100) comprises the audio signal normalization step (10,10', 10, 10', 10). Clause 11: The method (100) in accordance with any of the preceding clauses, wherein the step of determining (120) the audio patterns, determining (120) a window length and the windowing (140) are performed automatically and / or are performed by the use of artificial intelligence. Clause 12: The method (100) in accordance with clause 11, wherein the steps are performed by using a decision tree algorithm, a random forest algorithm, a naive Bayes algorithm, an Adaboost algorithm, an algorithm implemented by a neural network and / or a support vector machine algorithm. Clause 13: The method (100) in accordance with any of the preceding clauses, wherein the audio signal (10,10', 10, 10', 10j) is a recording of a sequence of a dog's heartbeat and / or a recording of a sequence of a dog's heart murmur. Clause 14: An apparatus (30) for analyzing (150) an acoustic signal (12a, 12b and 12c) having a time period (T0 to T6) and comprising a plurality of repeated audio patterns, the apparatus (30) comprising: an interface for receiving (110) the audio signal (10,10', 10, 10', 10) comprising the acoustic signal (12a, 12b and 12c); and a processor (32) configured to determine the repeated audio pattern within the acoustic signal (12a, 12b and 12c) and to determine a window length corresponding to a plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o), wherein the window length divides the time period (T0 to T6) of the acoustic signal (12a, 12b and 12c) into the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o); and to window the acoustic signal (12a, 12b and 12c) in order to obtain the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o). IVIA / a / ¿U¿ó / UU I Ί 40 Clause 15: A system for performing an analysis comprising the apparatus (30) according to clause 14 and a microphone (34) or preferably the apparatus (30) according to clause 14 and a stethoscope comprising a microphone (34) or more preferably the apparatus (30) according to clause 14 and a digital stethoscope comprising a microphone (34). Clause 16: A computer program having program code comprising instructions for executing method (100) in accordance with any of clauses 1 to 13. While some aspects have been described in the context of a device, it is clear that these aspects also represent a description of the corresponding method, in which a block or device corresponds to a step of the method or a feature of a step of the method. Similarly, the aspects described in the context of a step of the method also represent a description of a corresponding block or element or a feature of a corresponding device. Some or all of the steps of the method may be executed by (or using) a hardware device, such as a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, any one or more of the major steps of the method may be executed by such a device. Depending on certain implementation requirements, the embodiments of the invention can be implemented in hardware or software, or at least partially in software, or at least partially in hardware or software. The implementation can be carried out using a digital storage medium, for example, a floppy disk, DVD, Blu-ray disc, CD, ROM, PROM, EPROM, EEPROM, or FLASH memory, which has electronically readable control signals stored therein, cooperating (or capable of cooperating) with a programmable computer system in such a way as to execute the respective method. Therefore, the digital storage medium can be readable by a computer. Some embodiments according to the invention comprise a data carrier comprising electronically readable control signals, capable of cooperating with a programmable computer system in such a way as to execute one of the methods described herein. In general, the embodiments of the present invention can be implemented in the form of a computer program product with program code, where the program code fulfills the function of executing one of the methods when running the computer program on a computer. The program code can be stored, for example, on a machine-readable medium. IVIA / a / ¿U¿ó / UU I Ί 40 Other forms of implementation include the computer program to execute one of the methods described herein, stored on a machine-readable carrier. In other words, one form of implementation of the method of the invention consists, therefore, in a computer program consisting of program code for executing one of the methods described herein by running the computer program on a computer. A further embodiment of the methods of the invention therefore consists of a data carrier (or digital storage medium, or computer-readable medium) comprising, recorded thereon, the computer program for executing one of the methods described herein. The data carrier, the digital storage medium, or the recorded medium is generally tangible and / or non-transient. Another embodiment of the invention is, therefore, a data stream or signal sequence representing the computer program for executing one of the methods described herein. The data stream or signal sequence may be configured, for example, to be transmitted over a data communication connection, such as the internet. A further embodiment comprises a processing means, for example a computer, or a programmable logic device, configured or adapted to execute one of the methods described in this document. A further embodiment comprises a computer on which the computer program has been installed to execute one of the methods described herein. Another embodiment of the invention comprises an apparatus or system configured to transfer (for example, electronically or optically) a computer program for executing one of the methods described herein to a receiver. The receiver may be, for example, a computer, a mobile device, a memory device, or the like. The apparatus or system may comprise, for example, a file server for transferring a computer program to the receiver. In some embodiments, a programmable logic device (e.g., a field-programmable gate array) can be used to execute some or all of the functionalities of the methods described herein. In some embodiments, a field-programmable gate array can cooperate with a microprocessor to execute one of the methods described herein. Generally, the methods are preferably executed by any hardware device. The embodiments described above are merely illustrative of the principles of the present invention. It is understood that modifications and variations of the arrangements and details described herein should be obvious to other persons skilled in the art. Therefore, it is only intended to be limited by the scope of the ML / a / ZUZÓ / UU 1140 following patent claims and not by the specific details presented as a description and explanation of the embodiment of the present.

Claims

1. A method (100) for analyzing (150) an acoustic signal (12a, 12b and 12c) having a time period (T0 to T6) and comprising a plurality of repeated audio patterns, comprising the following steps: receiving (110) an audio signal (10,10', 10, 10', 10) comprising the acoustic signal (12a, 12b and 12c), wherein the audio signal (10,10', 10, 10', 10', 10) is a recording of a heartbeat sequence of an animal, preferably a non-human mammal, more preferably a dog, and / or a recording of a heart murmur sequence of an animal, preferably a non-human mammal, more preferably a dog; determining (120) the repeated audio patterns within the acoustic signal (12a, 12b and 12c);determine (120) a window length corresponding to a plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o), wherein the window length divides the time period (TO to T6) of the acoustic signal (12a, 12b and 12c) into the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o); wherein the determination (120) of the window length is performed separately for each window of the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o) and windowing (140) the acoustic signal to obtain the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o); wherein the step of determining (120) the audio patterns, determining (120) a window length and the windowing (140) step are executed automatically.

2. The method (100) according to claim 1, wherein the method (100) comprises the additional step of analyzing (150) the respective windows (14a, 14b, 14c, 14a'a 14o', 14aa 14o).

3. The method (100) according to claim 2, wherein the additional analysis step (150) comprises the step of performing a feature extraction to obtain one or more extracted features describing the respective pattern.

4. The method (100) according to claim 3, wherein the features to be extracted are from the group comprising the name feature, the time-domain feature, and / or the frequency-domain feature; and / or wherein the feature to be extracted is from the group comprising a maximum, a mean, median, standard deviation, variance, skewness, kurtosis, mean absolute deviation, 25th quantile, 75th quantile, entropy, zero-crossing rate, crest factor, duration of a first peak and / or second peak within the pattern, duration between the first peak and the second peak within the pattern, duration between the second peak of a first pattern and the first peak of a subsequent pattern, cepstral coefficients at the mel frequency, hue chroma, spectral flatness, spectral kurtosis, spectral skewness, spectral slope, dominant frequency, bandwidth, spectral centroid, spectral flux,spectral displacement, class information, severity information, position information, breed information, weight information, additional information and / or other parameters or another combination thereof and / or where the feature extraction step comprises the step of redefining the range of values ​​of one or more extracted features so that the range of values ​​corresponding to one or more extracted features is defined between a minimum value or 0 and a maximum value or 1.

5. The method (100) according to any of claims 2-4, wherein the method comprises the step of submitting a report on the analysis or wherein the method comprises the step of issuing a report on the analysis, the report comprising information on a disease or a breath of the animal, preferably the non-human mammal, more preferably the dog.

6. The method (100) according to any of the preceding claims, wherein the repeated audio patterns are identical to each other, substantially identical to each other, similar to each other, comprise one or more comparable peaks of the respective amplitude plotted over time and / or comprise one or more comparable peaks of the amplitude plotted over time and comparable amplitude values ​​at the respective time point within the length of the window.

7. The method (100) according to any of the preceding claims, wherein the length of the window is equal.

8. The method (100) according to any of the preceding claims, wherein the window length is determined as a function of the frequency of the repeating pattern. IVIA / a / ¿U¿ó / UU I Ί 40 9. The method (100) according to any of the preceding claims, wherein the method (100) further comprises the step of ignoring one or more windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o) without an audio pattern similar or equal to the plurality of repeated audio patterns.

10. The method (100) according to any of the preceding claims, wherein each repeated audio pattern is defined by one or more peaks; and / or wherein each repeated audio pattern is defined by one or more peaks in combination with a base level, wherein one or more peaks have an amplitude value that is at least five times greater than the base level; and / or wherein each repeated audio pattern is defined by a systole and / or diastole.

11. The method (100) according to any of the preceding claims, wherein the method (100) comprises the audio signal normalization step (10,10', 10, 10', 10).

12. The method (100) according to any of the preceding claims, wherein the step of determining (120) the audio patterns, determining (120) a window length and the windowing (140) are performed by using artificial intelligence.

13. The method (100) according to claim 12, wherein the steps are performed by using a decision tree algorithm, a random forest algorithm, a naive Bayes algorithm, an adaboost algorithm and / or a support vector machine algorithm.

14. The method (100) according to any of the preceding claims, wherein the acoustic signal (12a, 12b and 12c) comprises the heartbeat sequence of an animal, preferably a non-human mammal, more preferably a dog, wherein the heartbeat sequences comprise the plurality of repeated audio patterns.

15. The method (100) according to any of the preceding claims, wherein the determination (120) of the window length comprises determining a boundary between two repeated audio patterns in order to determine the window length corresponding to the respective window and / or IVIA / a / ¿U¿ó / UU I Ί 40 where the determination (120) of the window lengths comprises determining a characteristic feature, pulse, peak, pattern, systole and / or diastole of a window in order to determine the beginning of a window and to determine the respective characteristic, pulse, peak, pattern, systole and / or diastole of a subsequent window in order to determine the end of said window and / or wherein the determination (120) of the window lengths comprises determining the window lengths by determining a beginning and an end of a window.

16. The method (100) according to any of the preceding claims, wherein the window lengths vary over time and / or wherein the window lengths vary from a first window of the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o) to a subsequent window of the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o).

17. The method (100) according to any of the preceding claims, wherein the length of the windows varies according to the heart rate of the animal, preferably the non-human mammal, more preferably the dog; and / or wherein the method further comprises the step of determining the heart rate.

18. An apparatus (30) for analyzing (150) an acoustic signal (12a, 12b and 12c) having a time period (TO to T6) and comprising a plurality of repeated audio patterns, wherein the apparatus (30) comprises: an interface for receiving (110) the audio signal (10,10', 10, 10', 10) comprising the acoustic signal (12a, 12b and 12c);The audio signal (10, 10', 10, 10', 10) is a recording of a heartbeat sequence of an animal, preferably a non-human mammal, more preferably a dog, and / or a recording of a heart murmur sequence of an animal, preferably a non-human mammal, more preferably a dog, and a processor (32) that is configured to determine the repeated audio pattern within the acoustic signal (12a, 12b, and 12c) and to determine a window length corresponding to a plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o), wherein the window length divides the time period (T0 to T6) of the acoustic signal (12a, 12b, and 12c) into the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o). wherein the processor (32) determines the window length of each window of the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o) separately;and windowing the acoustic signal (12a, 12b and 12c) to obtain the plurality of windows (14a, 14b, 14c, 14a' to 14o', 14a to 14o); ML / a / ZUZÓ / UU I Ί 40 in which the step of determining (120) the audio patterns, determining (120) a window length and windowing (140) are performed automatically.; 19. A system for performing an analysis comprising the apparatus (30) according to claim 18 and 5, a microphone (34), or preferably the apparatus (30) according to claim 18 and a stethoscope comprising a microphone (34), or more preferably the apparatus (30) according to claim 18 and a digital stethoscope comprising a microphone (34).

20. A non-transient computer-readable means for analyzing (150) an acoustic signal (12a, 12b and 12c) having a time period (T0 to T6) and comprising a plurality of repeated audio patterns according to any of claims 1 to 17.

Images