Digital Audio Signal Research Articles

Optimizing poultry audio signal classification with deep learning and burn layer fusion

This study introduces a novel deep learning-based approach for classifying poultry audio signals, incorporating a custom Burn Layer to enhance model robustness. The methodology integrates digital audio signal processing, convolutional neural networks (CNNs), and the innovative Burn Layer, which injects controlled random noise during training to reinforce the model's resilience to input signal variations. The proposed architecture is streamlined, with convolutional blocks, densely connected layers, dropout, and an additional Burn Layer to fortify robustness. The model demonstrates efficiency by reducing trainable parameters to 191,235, compared to traditional architectures with over 1.7 million parameters. The proposed model utilizes a Burn Layer with burn intensity as a parameter and an Adamax optimizer to optimize and address the overfitting problem. Thorough evaluation using six standard classification metrics showcases the model's superior performance, achieving exceptional sensitivity (96.77%), specificity (100.00%), precision (100.00%), negative predictive value (NPV) (95.00%), accuracy (98.55%), F1 score (98.36%), and Matthew’s correlation coefficient (MCC) (95.88%). This research contributes valuable insights into the fields of audio signal processing, animal health monitoring, and robust deep-learning classification systems. The proposed model presents a systematic approach for developing and evaluating a deep learning-based poultry audio classification system. It processes raw audio data and labels to generate digital representations, utilizes a Burn Layer for training variability, and constructs a CNN model with convolutional blocks, pooling, and dense layers. The model is optimized using the Adamax algorithm and trained with data augmentation and early-stopping techniques. Rigorous assessment on a test dataset using standard metrics demonstrates the model's robustness and efficiency, with the potential to significantly advance animal health monitoring and disease detection through audio signal analysis.

Audio signal processing technology in Arduino controlling

In recent years, the application of audio signal processing technology is frequent, which is beneficial for developing both technologies. A significant advantage of the application of audio signal processing is allowing users to input their commands by voice, which is why it is widely used in Arduino controlling. Besides, Arduino can also be applied in audio signal processing. Because Arduino is an open-source multi-functions platform with the features of low cost, its utilization will be wider and more flexible in audio-digital signal processing and many other fields in the following time. So, the combination of these two technologies can produce more and more meaningful outcomes. This paper first introduces the technical features and application scenarios of these two technologies respectively, and then analyzes some samples of the combination of the two technologies specifically and summarizes them in different aspects to know about the current situation of these fields. By deeply understanding the inadequacies of these projects, this paper also analyzes the future developing trends of the fields that can solve the current problems and improve the quality of the projects.

Software-architectural configuration of the multifunctional audio digital signal processor module for signal mediatesting of audio devices

Objectives. The aim of this study is to develop and analyze parameters for a multifunctional audio module based on the ADAU1701 audio digital signal processor in the SigmaStudio environment. This will be used for testing audio devices in the following modes: routing of balanced and unbalanced audio channels according to the differential scheme Di-Box/R Di-Box; spatiotemporal and dynamic audio processing; three-band monochannel cross-separation with independent equalization; and correction of the frequency response of the audio channel with tracking notch auto-suppression of electro-acoustic positive feedback in a given spectral band.Methods. Visual-graphical architectural programming of audio modules in the SigmaStudio and Flowstone, as well as algorithms for real-time signal audio measurements and analysis of experimental data in the REW and Soundcard Oscilloscope are used.Results. The characteristics of the Di-Box/R Di-Box circuit were studied, in order to estimate the effect of differential signal conversion on the signal-to-noise ratio in the audio signal path. The characteristics of the reverberation and saturation submodules were established. Furthermore, the effect of equalization modes on the frequency response correction of a studio audio monitor was determined. The paper also studied the effect of an audio compressor on the dynamic range and the level of the output signal. The experimental results of the submodule for compensating the frequency response of an audio monitor using matched filtering were established, and the spectral characteristics of the submodule for automatic suppression of electro-acoustic positive feedback were obtained.Conclusions. The software architecture of a multifunctional audio module based on the ADAU1701 audio digital signal processor for testing and debugging media devices in a given spectral-dynamic and spectral-temporal ranges was designed. Balanced routing allows the effect of noise induced into the audio channel to be reduced 20-fold, thus enabling calibration of pickup audio devices. The audio signal processing submodule provides: compression response in the dynamic range from −27 to 18.6 dB with the possibility of equalization parameterization in the range of 0.04–18 kHz; reverberation response in the range from 0.5–3000 ms; audio-channel cross-division into 3 with the ability to adjust the amplitude-frequency response in the dynamic range from −30 to 30 dB. The auto-correction submodule of the amplitude-frequency response allows the dynamic nonuniformity of the amplitude-frequency response to be reduced by 40 dB. The auto-suppression submodule of electro-acoustic positive feedback provides notch formant suppression up to −100 dB with an input dynamic range from −50 to 80 dB.

Robust Watermarking Using FFT and Cordic QR Techniques

Digital media sharing and access in today’s world of the internet is very frequent for every user. The management of digital rights may come into threat easily as the accessibility of data through the internet become wide. Sharing digital information under security procedures can be easily compromised due to the various vulnerabilities floating over the internet. Existing research has been tied to protecting internet channels to ensure the safety of digital data. Researchers have investigated various encryption techniques to prevent digital rights management but certain challenges including external potential attacks cannot be avoided that may give unauthorized access to digital media. The proposed model endorsed the concept of watermarking in digital data to uplift media security and ensure digital rights management. The system provides an efficient procedure to conduct over-watermarking in digital audio signals and confirm the avoidance of ownership of the host data. The proposed technique uses a watermark picture as a signature that has been initially encrypted with Arnold's cat map and cyclic encoding before being embedded. The upper triangular R-matrix component of the energy band was then created by using the Fast Fourier transform and Cordic QR procedures to the host audio stream. Using PN random sequences, the encrypted watermarking image has been embedded in the host audio component of the R-matrix. The same procedure has been applied to extract the watermark image from the watermarked audio. The proposed model evaluates the quality of the watermarked audio and extracted watermark image. The average PSNR of the watermarked audio is found to be 37.01 dB. It has also been seen that the average PSNR, Normal cross-correlation, BER, SSMI (structure similarity index matric) value for the extracted watermark image is found to be 96.30 dB, 0.9042 units, 0.1033 units, and 0.9836 units respectively. Further, the model has been tested using various attacks to check its robustness. After applying attacks such as noising, filtering, cropping, and resampling on the watermarked audio, the watermark image has been extricated and its quality has been checked under the standard parameters. It has been found that the quality of the recovered watermark image satisfying enough to justify the digital ownership of the host audio. Hence, the proposed watermarking model attains a perfect balance between imperceptibility, payload, and robustness.

Imperceptible Technique to Secure Digital Audio Signals while Attaining Trade-Off among the Parameter of Magic Triangle

Abstract: This paper outlines an audio encryption method that is used during audio compression. Security for the multimedia data is provided by encryption methods. Here, audio encryption is a method used to communicate secret information by subtly altering an audio stream. It is a technique that guarantees secure data transfer between parties that often interact online. Here, we offer a fresh method for dealing with issues raised by audio encryption's substitution mechanism. We employ an enhanced RSA encryption technique, which is incredibly difficult to crack, to encrypt messages in the first level of protection. The encrypted message bits are embedded into random LSB layers to boost robustness against purposeful assaults, in which the hackers constantly attempt to reveal the hidden message, as well as some inadvertent attacks, like noise addition. Here, GA operators are utilised to lessen distortion. The major goal of this work is to propose a good, effective way for hiding data from hackers and sending it to the target in a safer manner by maintaining randomization in message bit insertion into audio data.

Behavioral and Electrical Modeling of a 0.5-V Third-Order Continuous-Time Sigma-Delta Modulator with FIR DAC for Audio Applications

Most mobile and wearable devices present digital audio signal processing capabilities. Since the nature of audio signals is analog, there is a need to use analog-to-digital converters (ADCs) with high-resolution for a high signal-to-noise ratio audio acquisition. This paper presents the high-level modeling and design of a continuous-time third-order sigma-delta modulator (CT-SDM) with an FIR DAC for audio devices, using a supply voltage of 0.5 V. The design is divided in three steps and is carried out using the Delta-sigma toolbox and a discrete-time to continuous-time (DT-CT) transformation. First, the schematic implementation with verilogA models is done to estimate the first-integrator amplifier specifications for the modulator to provide 14 bits of ENOB. Following, a two-stage inverter-based amplifier is designed and used to verify the design strategy. Finally, a transistor-level implementation of OTAs and comparator is done to evaluate the CT-SDM performance. An in-depth analysis and discussion are presented to explain the achieved results with those transistor-level circuits.

Wave Digital Models of Piezoelectric Transducers for Audio Applications

The number of consumer electronics devices integrating piezoelectric (PE) transducers as flat-panel loudspeakers has recently experienced a steep increase. Being very thin, in fact, PE transducers well cope with the miniaturization process that characterizes the market. However, at the same time, their reduced dimensions cause the sound pressure level to be poor at low frequencies, impairing the overall acoustic response. In this article, we derive novel efficient discrete-time models of PE transducers in the wave digital (WD) domain such that they can be integrated in the future into signal processing algorithms for audio output enhancement. WD filters are, in fact, making headway among audio digital signal processing techniques based on circuit equivalent models thanks to their efficiency, robustness, and accuracy. Starting from circuital representations of the piezo constitutive equations, we derive both lumped and distributed models. In particular, we show how it is possible to implement the frequency-dependent elements characterizing Mason’s model in the WD domain, contrary to what can be done using mainstream SPICE-like simulators. Such WD implementations come in handy for the simulation and fast prototyping of PE systems, paving the way toward the design of model-based digital signal processing algorithms for enhancing the transducer acoustic performance.

Noise-Immune Marking of Digital Audio Signals in Audio Stegosystems with Multiple Inputs and Multiple Outputs

Noise-Immune Marking of Digital Audio Signals in Audio Stegosystems with Multiple Inputs and Multiple Outputs

Discrete Wavelet Transform in digital audio signal processing: A case study of programming languages performance analysis

The time spent processing the test set is a crucial factor in research projects involving audio signal transforms. The present work analyzes the performance of a set of programming languages commonly used in this type of scientific research. The results showed the C and C++ languages as faster in a test suite with 8 different programming languages implementing the discrete Wavelet transform.

Audio Segmentation Techniques and Applications Based on Deep Learning

Audio processing has become an inseparable part of modern applications in domains ranging from health care to speech-controlled devices. In automated audio segmentation, deep learning plays a vital role. In this article, we are discussing audio segmentation based on deep learning. Audio segmentation divides the digital audio signal into a sequence of segments or frames and then classifies these into various classes such as speech recognition, music, or noise. Segmentation plays an important role in audio signal processing. The most important aspect is to secure a large amount of high-quality data when training a deep learning network. In this study, various application areas, citation records, documents published year-wise, and source-wise analysis are computed using Scopus and Web of Science (WoS) databases. The analysis presented in this paper supports and establishes the significance of the deep learning techniques in audio segmentation.

A New High-Capacity Audio Watermarking Based on Wavelet Transform using the Golden Ratio and TLBO Algorithm

Digital watermarking is one of the best solutions for copyright infringement, copying, data verification, and illegal distribution of digital media. Recently, the protection of digital audio signals has received much attention as one of the fascinating topics for researchers and scholars. In this paper, we presented a new high-capacity, clear, and robust audio signaling scheme based on the DWT conversion synergy and golden ratio advantages using the TLBO algorithm. We used the TLBO algorithm to determine the effective frame length and embedded range, and the golden ratio to determine the appropriate embedded locations for each frame. First, the main audio signal was broken down into several sub-bands using a DWT in a specific frequency range. Since the human auditory system is not sensitive to changes in high-frequency bands, to increase the clarity and capacity of these sub-bands to embed bits we used the watermark signal. Moreover, to increase the resistance to common attacks, we framed the high-frequency bandwidth and then used the average of the frames as a key value. Our main idea was to embed an 8-bit signal simultaneously in the host signal. Experimental results showed that the proposed method is free from significant noticeable distortion (SNR about 29.68dB) and increases the resistance to common signal processing attacks such as high pass filter, echo, resampling, MPEG (MP3), etc.

A New Model to Detect COVID-19 Coughing and Breathing Sound Symptoms Classification from CQT and Mel Spectrogram Image Representation using Deep Learning

Deep Learning is a relatively new Artificial Intelligence technique that has shown to be extremely effective in a variety of fields. Image categorization and also the identification of artefacts in images are being employed in visual recognition. The goal of this study is to recognize COVID-19 artefacts like cough and also breath noises in signals from real-world situations. The suggested strategy considers two major steps. The first step is a signal-to-image translation that is aided by the Constant-Q Transform (CQT) and a Mel-scale spectrogram method. Next, nine deep transfer models (GoogleNet, ResNet18/34/50/100/101, SqueezeNet, MobileNetv2, and NasNetmobile) are used to extract and also categorise features. The digital audio signal will be represented by the recorded voice. The CQT will transform a time-domain audio input to a frequency-domain signal. To produce a spectrogram, the frequency will really be converted to a log scale as well as the colour dimension will be converted to decibels. To construct a Mel spectrogram, the spectrogram will indeed be translated onto a Mel scale. The dataset contains information from over 1,600 people from all over the world (1185 men as well as 415 women). The suggested DL model takes as input the CQT as well as Mel-scale spectrograms derived from the breathing and coughing tones of patients diagnosed using the coswara-combined dataset. With the better classification performance employing cough sound CQT and a Mel-spectrogram image, the current proposal outperformed the other nine CNN networks. For patients diagnosed, the accuracy, sensitivity, as well as specificity were 98.9%, 97.3%, and 98.1%, respectively. The Resnet18 is the most reliable network for symptomatic patients using cough and breath sounds. When applied to the Coswara dataset, we discovered that the suggested model's accuracy (98.7%) outperforms the state-of-the-art models (85.6%, 72.9%, 87.1%, and 91.4%) according to the SGDM optimizer. Finally, the research is compared to a comparable investigation. The suggested model is more stable and reliable than any present model. Cough and breathing research precision are good enough just to test extrapolation as well as generalization abilities. As a result, sufferers at their headquarters may utilise this novel method as a main screening tool to try and identify COVID-19 by prioritising patients' RT-PCR testing and decreasing the chance of disease transmission.

A Low-Cost, High-Performance Digital Signal Processor (DSP) for Audio Applications

Digital audio signal processing has become increasingly common in multimedia systems as a result of the development of general-purpose digital signal processors (DSP processors) and high-precision oversampling analog-to-digital (A/D) converters. Nevertheless, the solution based on standard DSP chips could be too expensive and power-hungry for portable and home entertainment audio systems. As a result, it is necessary to locate an alternative solution that is both low-cost and low-power. In the context of digital audio applications, this paper will describe how a DSP (Digital Signal Processor) is created, as well as how its architecture functions. The proposed digital signal processor (DSP) has six stages of pipeline with a fixed 24-bit data format, and each stage has 125 instructions. The instructions for the audio signal processing are provided in a very specific manner. Each and every command is handled within the confines of a single cycle.

A Method for Watermark Detection in Digital Audio Signals by Authorized Users

A Method for Watermark Detection in Digital Audio Signals by Authorized Users

A flexible method to create wave file features

Digital audio signal is one of the most important data type at present, it is used in various vital applications, such as human knowledge, security and banking applications, most applications require signal identification and recognition, and to increase the efficiency of these applications we must seek a method to represent the audio file by a small set of values called a features vector. In this paper research we will introduce an enhanced method of features extraction based on k-mean clustering. The method will be tested and implemented to show how the proposed method can reduce the efforts of voice identification, and can minimize the recognition time a set of voice extracted features must be used instead of using the voice wave file.

Fabrication and Characterization of Dual Functional InGaN LED Arrays as the Optical Transmitter and Receiver for Optical Wireless Communications

The feasibility of using InGaN/GaN multiple-quantum-well light-emitting diode arrays (LED arrays) as photodiodes (PDs) is investigated experimentally in addition to their light emitting function. Two discrete LED arrays are produced from one 4 × 4 LED array with a parallel-connected pixel configuration. Such compact designs are useful for light emission or detection at the transmitting/receiving terminals of optical wireless communication systems. Despite 4 × 2 LED arrays achieving a light output power of 67.4 mW at 250 mA, they exhibit an optical responsivity (detectivity) of 0.183 A/W (1.61 × 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">12</sup> cm Hz <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1/2</sup> W <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1</sup> ) under ultraviolet light illumination (λ = 380 nm) at zero bias. For 4 × 2 LED arrays, the presence of an appreciable ultraviolet light response, together with a high 3-dB bandwidth (~8 MHz) for modulated light detection, allowed us to build a 15 Mbit/s directed optical link with these LEDs functioning as both the optical transmitter and the receiver. Finally, the unitary LED array-based optical link is capable of real-time transmission of digital audio signals (data rate = 6 Mbit/s) at a propagation distance of 100 cm in free space even though some of the constituent pixels are inactive for light detection.

Bit Error Rate Performance of Ofdm Utilising Differentially Encoded 16 Star Qam with Differentially Coherent Demodulation in Awgn, Frequency Flat and Two-Path Fading Channels.(Dept.E)

The multicarrier transmission technique known as Orthogonal Frequency Division Multiplexing (OFDM) was devised in the 1960's for voiceband data transmission. Today there are two principles FDM applications, one is for the high speed digital subscriber loop and the other is for the broadcasting of digital audio and video signals. It will consider the use of OFDM for high-bit rate wireless applications. In order to improve spectral efficiency, 16 star quadrature amplitude modulation (QAM) is employed to modulate the parallel carriers in place of the usual quadrature phase shift -keying (QPSK). This paper investigates the bit error rate (BER) performance of OFDM utilising differentially encoded (DE) 16 star QAM and differentially coherent demodulation in frequency flat and two-path fading channels via Monte Carlo simulation. The BER performance of OFDM is also compared with equivalent single carrier systems.

Using Faust DSL to Develop Custom, Sample Accurate DSP Code and Audio Plugins for the Web Browser

The development and porting of virtual instruments or audio effects on the Web is a hot topic. Several initiatives are emerging, from industry-driven ones (e.g., Propellerhead Rack Extension running on the Web2), to more community based open-source projects. Most of them aim at adapting existing code bases (usually developed in native languages like C/C++) as well as facilitating the use of existing audio Digital Signal Processing (DSP) languages and platforms. Our two teams previously presented an open format for WebAudio Plugins coined WAP [2]. It aims at: (i) improving the interoperability of audio/MIDI plugins developed using pure Web APIs, (ii) porting existing native code bases, or (iii) using Domain Specific Languages (DSL). In this paper, we present a solution based around FAUST DSL, its Web-based editor, and the integration of a plugin GUI editor allowing to directly test, generate and deploy WAP plugins. We also evoke our collaborative work: one team hatching and improving FAUST, the other working on the recreation of tube guitar amplifiers and pedalboards within Web browsers. So as to fully illustrate the FAUST online framework, a case study is detailed with complete workflow, from the FAUST DSP source code written and tested in a fully functional online editor, to a self-contained plugin running in a separate host application.

CLSTM: Deep Feature-Based Speech Emotion Recognition Using the Hierarchical ConvLSTM Network

Artificial intelligence, deep learning, and machine learning are dominant sources to use in order to make a system smarter. Nowadays, the smart speech emotion recognition (SER) system is a basic necessity and an emerging research area of digital audio signal processing. However, SER plays an important role with many applications that are related to human–computer interactions (HCI). The existing state-of-the-art SER system has a quite low prediction performance, which needs improvement in order to make it feasible for the real-time commercial applications. The key reason for the low accuracy and the poor prediction rate is the scarceness of the data and a model configuration, which is the most challenging task to build a robust machine learning technique. In this paper, we addressed the limitations of the existing SER systems and proposed a unique artificial intelligence (AI) based system structure for the SER that utilizes the hierarchical blocks of the convolutional long short-term memory (ConvLSTM) with sequence learning. We designed four blocks of ConvLSTM, which is called the local features learning block (LFLB), in order to extract the local emotional features in a hierarchical correlation. The ConvLSTM layers are adopted for input-to-state and state-to-state transition in order to extract the spatial cues by utilizing the convolution operations. We placed four LFLBs in order to extract the spatiotemporal cues in the hierarchical correlational form speech signals using the residual learning strategy. Furthermore, we utilized a novel sequence learning strategy in order to extract the global information and adaptively adjust the relevant global feature weights according to the correlation of the input features. Finally, we used the center loss function with the softmax loss in order to produce the probability of the classes. The center loss increases the final classification results and ensures an accurate prediction as well as shows a conspicuous role in the whole proposed SER scheme. We tested the proposed system over two standard, interactive emotional dyadic motion capture (IEMOCAP) and ryerson audio visual database of emotional speech and song (RAVDESS) speech corpora, and obtained a 75% and an 80% recognition rate, respectively.

High-SNR steganography for digital audio signal in the wavelet domain

Imperceptible, robust, and embedding capacity are three main requirements for the steganography of digital audio signal. To enhance them, this study presents a novel steganography for digital audio signal in the wavelet domain. Since the performance of imperceptible and robust are usually in term of signal-to-noise ratio (SNR) and bit-error-rate (BER), we propose a quantization-based optimization model to maximize SNR and reduce BER in embedding secret message. In the proposed model, quantization technique with unknow coefficients of discrete wavelet transform (DWT) is rewritten as the first constraint. The adjustment of scaling DWT coefficients is considered as the second constraint. At the same time, signal-to-noise ratio (SNR) is converted into a performance index. In solving this model, we use matrix operations and Lagrange multiplier to obtain optimal DWT coefficients and scaling factors. Moreover, the invariant feature of the scaling factors against amplitude scaling attack is proved. In extraction, secret message can be detected without original audio signal. Experimental results show that the proposed steganography has high SNR and strong robustness against many malicious attacks when comparing to some exiting methods.