Sound Pattern Recognition Research Articles

Study on weak sound signal separation and pattern recognition under strong background noise in marine engineering

The extraction of weak acoustic signals under strong background noise is of great significance in the applications of target identification and localization. In this paper, the pulse signal with high randomness is set as the weak signal sound source, random noise and sine sound are used as the background noise. Under the condition of a signal-to-noise ratio of −20 dB, combined with blind source separation and neural network methods, the collected observation signals are subjected to weak sound signal separation and recognition research. The optimization method of centralization and scaling processing is used to eliminate the unfavorable influence of the uncertainty of the separated signal amplitude caused by the blind source separation method on the pattern recognition. The recognition result is verified by the combination of “weak impulse acoustic signal” and “random noise signal,” and the output vector (0.99 0.01 0.01) approaches (1 0 0), which is recognized as impulse acoustic signal. By combining blind source separation and neural network methods, the separation and identification of weak pulse signals under the condition of a signal-to-noise ratio of −20 dB can be achieved.

PATTERN RECOGNITION FROM IMAGES AND VIDEOS USING MACHINE LEARNING

Pattern recognition is a data analysis method that uses machine learning algorithms to automatically recognize patterns and regularities in data. This data can be anything from text and images to sounds or other definable qualities. Pattern recognition is supervised or unsupervised classification. Among the various frameworks in which pattern recognition has been traditionally formulated, the statistical approach structural and new has been most intensively studied and used in practice. More recently, neural network techniques and methods imported from statistical learning theory have deserved increasing attention. There are three main types of pattern recognition, dependent on the mechanism used for classifying the input data. Those types are: statistical, structural (or syntactic), and neural. Based on the type of processed data, it can be divided into image, sound, voice, and speech pattern recognition.

Remote-Control Sound Pattern Recognition for Auditory Assessment and Therapy in Smart Healthcare

A remote-control method is proposed for healthcare services involving auditory assessment and therapy through video-conferencing software. Different video-conferencing software platforms have different sound responses, and the user computer also has variations in sound performance because of its hardware and software configuration. These restrictions affect the ability of video-conferencing software to deliver health services that require accurate, reliable, and repeatable sound. In this paper, a remote-control system board is developed wherein sound for auditory assessment or therapy is stored. A low-cost processor is used to remotely deliver sound by sound pattern recognition. Service providers play a sound pattern locally to remotely control the system board. The desired audio files can be delivered to the client side without suffering from sound loss in terms of frequency range and loudness. Experiments were also conducted to verify the performance of the proposed method. The results show that the remote-control method for audio playback can be realized with a low-cost processor, such as a microcontroller unit for sound pattern recognition and that it accurately delivers audio files in terms of frequency and amplitude.

AI Sound Recognition on Asthma Medication Adherence: Evaluation With the RDA Benchmark Suite

Asthma is a common, usually long-term respiratory disease with negative impact on global society and economy. Treatment involves using medical devices (inhalers) that distribute medication to the airways and its efficiency depends on the precision of the inhalation technique. There is a clinical need for objective methods to assess the inhalation technique, during clinical consultation. Integrated health monitoring systems, equipped with sensors, enable the recognition of drug actuation, embedded with sound signal detection, analysis and identification from intelligent structures, that could provide powerful tools for reliable content management. Health monitoring systems equipped with sensors, embedded with sound signal detection, enable the recognition of drug actuation and could be used for effective audio content analysis. This paper revisits sound pattern recognition with machine learning techniques for asthma medication adherence assessment and presents the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Respiratory and Drug Actuation (RDA)</i> Suite <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>a</i></sup> for benchmarking and further research. The RDA Suite includes a set of tools for audio processing, feature extraction and classification procedures and is provided along with a dataset, consisting of respiratory and drug actuation sounds. The classification models in RDA are implemented based on conventional and advanced machine learning and deep networks’ architectures. This study provides a comparative evaluation of the implemented approaches, examines potential improvements and discusses on challenges and future tendencies.

Behind the mask(ing): how frogs cope with noise.

Albert Feng was a pioneer in the field of auditory neuroethology who used frogs to investigate the neural basis of spectral and temporal processing and directional hearing. Among his many contributions was connecting neural mechanisms for sound pattern recognition and localization to the problems of auditory masking that frogs encounter when communicating in noisy, real-world environments. Feng's neurophysiological studies of auditory processing foreshadowed and inspired subsequent behavioral investigations of auditory masking in frogs. For frogs, vocal communication frequently occurs in breeding choruses, where males form dense aggregations and produce loud species-specific advertisement calls to attract potential mates and repel competitive rivals. In this review, we aim to highlight how Feng's research advanced our understanding of how frogs cope with noise. We structure our narrative around three themes woven throughout Feng's research-spectral, temporal, and directional processing-to illustrate how frogs can mitigate problems of auditory masking by exploiting frequency separation between signals and noise, temporal fluctuations in noise amplitude, and spatial separation between signals and noise. We conclude by proposing future research that would build on Feng's considerable legacy to advance our understanding of hearing and sound communication in frogs and other vertebrates.

Effect of musical training on auditory memory

Abstract Background Musical practice leads to cognitive development, requiring the activation of several processing mechanisms that are based on encoding and storing sounds through memory. The objective was to evaluate the auditory memory (ability to recognize tonal patterns) in musicians and non-musicians, comparing their performance. Methods Two groups with normal hearing and aged between 17 and 18 years old were compared, one of musicians (N = 14, with average of 8 years of musical training) and a control group (N = 14). Pure tone audiogram and an auditory working memory/pattern recognition test of non-verbal sounds (tonal) were performed. The participant heard 2 consecutive sequences of 6 tonal sounds each and had to say whether they were the same or different from each other. In total 30 pairs of sequences are presented (binaurally). Results The mean of percentage of correct responses in the memory/pattern recognition test was higher in musician’s group than in control group with statistically significant differences between groups (P = 0.002). The difference between the mean of percentage of correct responses in each group was 11.44%. Conclusions The musicians present better auditory memory/recognition of non-verbal (tonal) sounds in relation to the non-musicians, which points to the effectiveness of musical practice in the cognitive development of working memory and recognition of sound patterns. This study, according to others in the field, shows that musical training improves memory and auditory skills, which are known to be fundamental to the academic success of young people in this age group.

Development of Delaminate Inspection by Using Sound Sensor and Pattern Recognition Technique

Windsurf manufacturing and inspection process have high influence for product quality. Recently windsurf manufacturer faced high nonconformity cost with delaminate problem. Current inspection process is performed by inspector skill which depends on the experience. The inspection method is proposed by using sound sensor and applying the Fast Fourier Transform to monitor the ranges of laminate and delaminate frequency. The pattern recognition technique is purposed to classify noise frequency in order to identify the laminate windsurf from the delaminate ones. This research investigates the delaminate among the polyvinyl chloride sheet, the biaxial carbon reinforced fabric and the fiber glass reinforced fabric. The area of laminate frequency in frequency domain shows the laminate point has high power spectrum density than the one from the delaminate point. The algorithm to detect the delaminate has been developed and proved by real windsurf. It’s proved that the method proposed can be identified the delaminate windsurf.

Analisis Bentuk Pola Suara Menggunakan Ekstraksi Ciri Mel-Frequencey Cepstral Coefficients (MFCC)

Sound is a part of the human body that is unique and can be distinguished, so its application can be used in sound pattern recognition technology, one of which is used for sound biometrics. This study discusses the analysis of the form of a sound pattern that aims to determine the shape of the sound pattern of a person's character based on the spoken voice input. This study discusses the analysis of the form of a sound pattern that aims to determine the shape of the sound pattern of a person's character based on the spoken voice input. This study uses the Melf-Frequency Cepstrum Coefficients (MFCC) method for feature extraction process from speaker speech signals. The MFCC process will convert the sound signal into several feature vectors which will then be displayed in graphical form. Analysis and design of sound patterns using Matlab 2017a software. Tests were carried out on 5 users consisting of 3 men and 2 women, each user said 1 predetermined "LOGIN" word, which for 15 words said. The results of the test are the form of a sound pattern between the characteristics of 1 user with other users. Keywords—Voice, Pattern, Feature Extraction, MFCC

The Effects of Static and Dynamic Visual Representations as Aids for Primary School Children in Tasks of Auditory Discrimination of Sound Patterns. An Intervention-based Study.

It has been proposed that non-conventional presentations of visual information could be very useful as a scaffolding strategy in the learning of Western music notation. As a result, this study has attempted to determine if there is any effect of static and dynamic presentation modes of visual information in the recognition of sound patterns. An intervention-based quasi-experimental design was adopted with two groups of fifth-grade students in a Spanish city. Students did tasks involving discrimination, auditory recognition and symbolic association of the sound patterns with non-musical representations, either static images (S group), or dynamic images (D group). The results showed neither statistically significant differences in the scores of D and S, nor influence of the covariates on the dependent variable, although statistically significant intra-group differences were found for both groups. This suggests that both types of graphic formats could be effective as digital learning mediators in the learning of Western musical notation.

How regularity representations of short sound patterns that are based on relative or absolute pitch information establish over time: An EEG study.

The recognition of sound patterns in speech or music (e.g., a melody that is played in different keys) requires knowledge about pitch relations between successive sounds. We investigated the formation of regularity representations for sound patterns in an event-related potential (ERP) study. A pattern, which consisted of six concatenated 50 ms tone segments differing in fundamental frequency, was presented 1, 2, 3, 6, or 12 times and then replaced by another pattern by randomly changing the pitch of the tonal segments (roving standard paradigm). In an absolute repetition condition, patterns were repeated identically, whereas in a transposed condition, only the pitch relations of the tonal segments of the patterns were repeated, while the entire patterns were shifted up or down in pitch. During ERP measurement participants were not informed about the pattern repetition rule, but were instructed to discriminate rarely occurring targets of lower or higher sound intensity. EPRs for pattern changes (mismatch negativity, MMN; and P3a) and for pattern repetitions (repetition positivity, RP) revealed that the auditory system is able to rapidly extract regularities from unfamiliar complex sound patterns even when absolute pitch varies. Yet, enhanced RP and P3a amplitudes, and improved behavioral performance measured in a post-hoc test, in the absolute as compared with the transposed condition suggest that it is more difficult to encode patterns without absolute pitch information. This is explained by dissociable processing of standards and deviants as well as a back propagation mechanism to early sensory processing stages, which is effective after less repetitions of a standard stimulus for absolute pitch.

Sound and vibration-based pattern recognition for wind turbines driving mechanisms

This paper proposes a pattern recognition approach for a Fault Detection and Diagnosis (FDD) system based on the wavelet and the fast Fourier transform. Both techniques are developed in an experimental set that simulates the driving mechanisms housed in the nacelle of a wind turbine (WT) with results being validated in a real wind farm. After a first separate approach of the vibration harmonics and the sound energy, the root mean square error (RMSE) is used to fuse the data into a common pattern. The pattern reveals accurate information for unstable features (e.g. the case of the sound) related to misalignments among other failures. Comparing the experiments with the pattern, it is observed that the pattern is often close to the induced failures with minor exceptions. Relations among all the measured points are also found. The usefulness of the findings lies in the possibility of monitoring inaccessible devices considering this relation. Cost savings based on the strategic placement of the sensors can be intended too. The FDD will ensure the implementation of predictive actions before the occurrence of a catastrophic failure in an area where there is an ongoing challenge for being competitive.

音響パターン認識を用いた打音識別システムの研究

音響パターン認識を用いた打音識別システムの研究

A recognition method research based on the heart sound texture map

In order to improve the Heart Sound recognition rate and reduce the recognition time, in this paper, we introduces a new method for Heart Sound pattern recognition by using Heart Sound Texture Map. Based on the Heart Sound model, we give the Heart Sound time-frequency diagram and the Heart Sound Texture Map definition, we study the structure of the Heart Sound Window Function principle and realization method, and then discusses how to use the Heart Sound Window Function and the Short-time Fourier Transform to obtain two-dimensional Heart Sound time-frequency diagram, propose corner correlation recognition algorithm based on the Heart Sound Texture Map according to the characteristics of Heart Sound. The simulation results show that the Heart Sound Window Function compared with the traditional window function makes the first (S1) and the second (S2) Heart Sound texture clearer. And the corner correlation recognition algorithm based on the Heart Sound Texture Map can significantly improve the recognition rate and reduce the expense, which is an effective Heart Sound recognition method.

Songbirds use spectral shape, not pitch, for sound pattern recognition

Humans easily recognize "transposed" musical melodies shifted up or down in log frequency. Surprisingly, songbirds seem to lack this capacity, although they can learn to recognize human melodies and use complex acoustic sequences for communication. Decades of research have led to the widespread belief that songbirds, unlike humans, are strongly biased to use absolute pitch (AP) in melody recognition. This work relies almost exclusively on acoustically simple stimuli that may belie sensitivities to more complex spectral features. Here, we investigate melody recognition in a species of songbird, the European Starling (Sturnus vulgaris), using tone sequences that vary in both pitch and timbre. We find that small manipulations altering either pitch or timbre independently can drive melody recognition to chance, suggesting that both percepts are poor descriptors of the perceptual cues used by birds for this task. Instead we show that melody recognition can generalize even in the absence of pitch, as long as the spectral shapes of the constituent tones are preserved. These results challenge conventional views regarding the use of pitch cues in nonhuman auditory sequence recognition.

An auditory feature detection circuit for sound pattern recognition.

From human language to birdsong and the chirps of insects, acoustic communication is based on amplitude and frequency modulation of sound signals. Whereas frequency processing starts at the level of the hearing organs, temporal features of the sound amplitude such as rhythms or pulse rates require processing by central auditory neurons. Besides several theoretical concepts, brain circuits that detect temporal features of a sound signal are poorly understood. We focused on acoustically communicating field crickets and show how five neurons in the brain of females form an auditory feature detector circuit for the pulse pattern of the male calling song. The processing is based on a coincidence detector mechanism that selectively responds when a direct neural response and an intrinsically delayed response to the sound pulses coincide. This circuit provides the basis for auditory mate recognition in field crickets and reveals a principal mechanism of sensory processing underlying the perception of temporal patterns.

The social weight of spoken words

Speech serves a linguistic function, cueing sounds and words, and a social function, cueing talkers and their social attributes. Listeners readily map sound patterns in speech to social representations. This mapping introduces social biases on the recognition and encoding of sound patterns produced by different groups and individuals.

Low-power wearable respiratory sound sensing.

Building upon the findings from the field of automated recognition of respiratory sound patterns, we propose a wearable wireless sensor implementing on-board respiratory sound acquisition and classification, to enable continuous monitoring of symptoms, such as asthmatic wheezing. Low-power consumption of such a sensor is required in order to achieve long autonomy. Considering that the power consumption of its radio is kept minimal if transmitting only upon (rare) occurrences of wheezing, we focus on optimizing the power consumption of the digital signal processor (DSP). Based on a comprehensive review of asthmatic wheeze detection algorithms, we analyze the computational complexity of common features drawn from short-time Fourier transform (STFT) and decision tree classification. Four algorithms were implemented on a low-power TMS320C5505 DSP. Their classification accuracies were evaluated on a dataset of prerecorded respiratory sounds in two operating scenarios of different detection fidelities. The execution times of all algorithms were measured. The best classification accuracy of over 92%, while occupying only 2.6% of the DSP's processing time, is obtained for the algorithm featuring the time-frequency tracking of shapes of crests originating from wheezing, with spectral features modeled using energy.

KOMUNIKATOR TUNARUNGU DAN TUNANETRA

This paper describes the design and implementation of the deaf and blind communicator with the aim that they can communicate by using a Braille Codes per character and speech recognition using a computer. Communicator device is designed using AT89C51 microcontroller to change the characters that is sent by the computer into Braille code. The signal transmitted from the computer by hearing impairment through RS-232 serial interface with the help of driver IC ULN2803 to drive the solenoid. The end of the solenoid form a pattern of Braille code. Speech pattern recognition used are dictation mode with discrete speech method, whole word, large vocabulary and speaker dependent is designed with SAPI 5.1, Microsoft Speech Engine SDK 5.1 and Delphi software 6.0 to create an application program. On testing who performed indicate that the all the characters that is sent from keyboard can be converted become Braille Character who represented through the solenoid. While the recognition of sound patterns of the microphone most of the well can be translated into the characters displayed on a computer monitor. The success rate in speech recognition can be influenced by several factors such as differences in the sound at the time of training and at the time of dictation, noise from the environment and the quality of the microphone being used.

Spectrogram Enhancement By Edge Detection Approach Applied To Bioacoustics Calls Classification

Accurate recognition of sound patterns in spectrograms is important step for further recognition applications. However, background noise forms fundamental problem regardless the species under study. In this paper, crest factor feature was extracted from the limited dynamic range spectrogram. The developed crest factor image behaved as smoothed version of the spectrogram, at which edges of the involved sound patterns were detected without the need of prior smoothing filters and their scaling constraints. Attached noise – surrounds the detected edges – was removed, to form the enhanced spectrogram. The method was compared to other enhancement approaches such like spectral Subtraction and wavelet packet decomposition. Comparison was performed on different structure patterns of bats and birds. Results indicate how the method is promising for efficiently enhancing the spectrogram while preserving its temporal and spectral accuracy. The method correctly classified three bioacoustics species with an accuracy of 94.59%, using few 2D features of their enhanced spectrograms

Supervised Classification of Basaltic Aggregate Particles Based on Texture Properties

AbstractThe strength and durability of materials produced from aggregates (e.g., concrete bricks, concrete, and ballast) are critically affected by the weathering of the particles, which is closely related to their mineral composition. It is possible to infer the degree of weathering from visual features derived from the surface of the aggregates. By using sound pattern recognition methods, this study shows that the characterization of the visual texture of particles, performed by using texture-related features of gray scale images, allows the effective differentiation between weathered and nonweathered aggregates. The selection of the most discriminative features is also performed by taking into account a feature ranking method. The evaluation of the methodology in the presence of noise suggests that it can be used in stone quarries for automatic detection of weathered materials.