Audio Authentication Research Articles

Estimating the first and second derivatives of discrete audio data

A new method for estimating the first and second derivatives of discrete audio signals intended to achieve higher computational precision in analyzing the performance and characteristics of digital audio systems is presented. The method could find numerous applications in modeling nonlinear audio circuit systems, e.g., for audio synthesis and creating audio effects, music recognition and classification, time-frequency analysis based on nonstationary audio signal decomposition, audio steganalysis and digital audio authentication or audio feature extraction methods. The proposed algorithm employs the ordinary 7 point-stencil central-difference formulas with improvements that minimize the round-off and truncation errors. This is achieved by treating the step size of numerical differentiation as a regularization parameter, which acts as a decision threshold in all calculations. This approach requires shifting discrete audio data by fractions of the initial sample rate, which was obtained by fractional delay FIR filters designed with modified 11-term cosine-sum windows for interpolation and shifting of audio signals. The maximum relative error in estimating first and second derivatives of discrete audio signals are respectively in order of 10-13\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$10^{-13}$$\\end{document} and 10-10\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$10^{-10}$$\\end{document} over the entire audio band, which is close to double-precision floating-point accuracy for the first and better than single-precision floating-point accuracy for the second derivative estimation. Numerical testing showed that this performance of the proposed method is not influenced by the type of signal being differentiated (either stationary or nonstationary), and provides better results than other known differentiation methods, in the audio band up to 21 kHz.

Audio-deepfake detection: Adversarial attacks and countermeasures

Audio has always been a powerful resource for biometric authentication: thus, numerous AI-based audio authentication systems (classifiers) have been proposed. While these classifiers are effective in identifying legitimate human-generated input their security, to the best of our knowledge, has not been explored thoroughly when confronted with advanced attacks that leverage AI-generated deepfake audio. This issue presents a serious concern regarding the security of these classifiers because, e.g., samples generated using adversarial attacks might fool such classifiers, resulting in incorrect classification. In this study, we prove the point: we demonstrate that state-of-the-art audio deepfake classifiers are vulnerable to adversarial attacks. In particular, we design two adversarial attacks on a state-of-the-art audio-deepfake classifier, i.e., the Deep4SNet classification model, which achieves 98.5% accuracy in detecting fake audio samples. The designed adversarial attacks11The code of the attacks will be released open-source in the camera ready. leverage a generative adversarial network architecture and reduce the detector’s accuracy to nearly 0%. In particular, under graybox attack scenarios, we demonstrate that when starting from random noise, we can reduce the accuracy of the state-of-the-art detector from 98.5% to only 0.08%. To mitigate the effect of adversarial attacks on audio-deepfake detectors, we propose a highly generalizable, lightweight, simple, and effective add-on defense mechanism that can be implemented in any audio-deepfake detector. Finally, we discuss promising research directions.

METADATA AND CONTAINER STRUCTURE ANALYSIS FOR AUDIO AUTHENTICATION

Audio recordings of crime scenes are frequently utilized as digital evidence due to the universal usage of mobile phones, but they must first be authenticated before being used as evidence in court. The technique of identifying whether an audio recording is real or has been altered or manipulated is known as audio authenticity. In order to investigate and validate audio, this study provides a cutting-edge authentication technique that can distinguish the difference between real and fraudulent audio. Herein, we analyzed audio recording samples from different mobile handsets like iPhone, Samsung, One Plus etc. It is important to note that the proposed method authenticates audio files without regard to the audio content, i.e., without concern to the speaker or the speech. The audio recordings were subjected to manipulation that a human cannot recognize them auditory. The present results show that it is possible to verify the authentication of audio recordings generated through mobile phones or any recorder using metadata and the container structure of the recorded and edited audio file

Mel spectrogram-based audio forgery detection using CNN

In this time of technology, digital speech can be created and falsified by a very diverse of hardware and software technologies. Audio copy-move forgery is an audio forgery technique that goals to create forged audio by hiding undesirable words or repeating wanted words in identical speech. Therefore, audio authentication has been a necessary requisition. In this study, an effective approach to spectral images based on audio copy-move forgery detection using convolutional neural networks (CNN) with data augmentation is proposed. There are only a few handcrafted methods conducted for the detection of audio copy-move forgery. None of the existing works on audio copy-move forgery detection has proposed deep feature learning from speech recording with Mel spectrogram. This is the first method to employ deep learning with Mel spectrogram of audio for the detection of audio copy-move forgery. The proposed CNN architecture classifies the suspicious Mel spectrogram images into two classes: original and forged. The proposed CNN system is successfully trained on these Mel spectrogram image feature extraction. The proposed algorithm has been tested on our datasets generated from Arabic Speech Corpus and TIMIT speech database. The results show the effectiveness, robustness of post-processing operations, and high accuracy of the proposed approach compared to other studies.

Shallow and deep feature fusion for digital audio tampering detection

Digital audio tampering detection can be used to verify the authenticity of digital audio. However, most current methods use standard electronic network frequency (ENF) databases for visual comparison analysis of ENF continuity of digital audio or perform feature extraction for classification by machine learning methods. ENF databases are usually tricky to obtain, visual methods have weak feature representation, and machine learning methods have more information loss in features, resulting in low detection accuracy. This paper proposes a fusion method of shallow and deep features to fully use ENF information by exploiting the complementary nature of features at different levels to more accurately describe the changes in inconsistency produced by tampering operations to raw digital audio. Firstly, the audio signal is band-pass filtered to obtain the ENF component. Then, the discrete Fourier transform (DFT) and Hilbert transform are performed to obtain the phase and instantaneous frequency of the ENF component. Secondly, the mean value of the sequence variation is used as the shallow feature; the feature matrix obtained by framing and reshaping of the ENF sequence is used as the input of the convolutional neural network; the characteristics of the fitted coefficients are obtained by curve fitting. Then, the local details of ENF are obtained from the feature matrix by the convolutional neural network, and the global information of ENF is obtained by fitting coefficient features through deep neural network (DNN). The depth features of ENF are composed of ENF global information and local information together. The shallow and deep features are fused using an attention mechanism to give greater weights to features useful for classification and suppress invalid features. Finally, the tampered audio is detected by downscaling and fitting with a DNN containing two fully connected layers, and classification is performed using a Softmax layer. The method achieves 97.03% accuracy on three classic databases: Carioca 1, Carioca 2, and New Spanish. In addition, we have achieved an accuracy of 88.31% on the newly constructed database GAUDI-DI. Experimental results show that the proposed method is superior to the state-of-the-art method.

Implementation of Graphical Password Authentication Technique for Security Using Cued Click Points Algorithm

Abstract: In today's world, most Internet applications still establish user authentication with a traditional text-based password. Designing a secure as well as user-friendly password-based method has long been on the agenda of security researchers. On the one hand, there are password manager programs that make it easy to create site-specific strong passwords from a single user's password to eliminate the memory burden caused by multiple passwords. We offer different levels of authentication such as Textual Authentication, Image Authentication and Audio Authentication to provide better security for applications. User will select username and password while registering in text step. During registration the user has to enter the registered username and password, if it matches the database then the user can log in to the system. In Image Authentication Model, we take image as input from user at the time of registration and put quid point, qud point is selected part of image which is selected by user. At the time of login the user has to select the image and select the part of the image which he/she wants to include at the time of registration, which is called as Cued Points. Keywords: Graphical password, Text based password, usability, security, Attacks, Authentication

Asymmetric Contrastive Learning for Audio Fingerprinting

Audio fingerprinting methods can compress audio contents into compact signatures so that we can save storage and reduce query time. This technology is widely used in many fields, such as audio retrieval, music information retrieval and audio authentication. However, most of the existing methods cannot balance the recognition accuracy, query speed and storage size well. This letter presents a novel self-supervised learning scheme called asymmetric contrastive learning to generate binary hash fingerprints of audio segments. Meanwhile, we design a new loss function named bidirectional asymmetric pairwise loss to minimize the loss of information. Experimental results show that our scheme can achieve a high top-1 hit rate on both music and speech datasets. Furthermore, the proposed scheme outperforms the previous work of real-value fingerprinting in query speed and storage size.

Research on Music Multiterminal Audio Authentication Based on Wireless Network

The current music multiterminal audio authentication algorithm does not consider the mutation of music signal, which leads to poor tamper detection ability and long time of audio authentication. By analyzing the characteristics and key technologies of wireless network, a wireless multiterminal audio system is established. The short‐term energy calculation method is used to consider the sudden change of music signal. The music signal is divided into note segments, and chroma features of half order notes are extracted. The robust hash value is calculated by nonuniform quantization method. The dynamic time warping algorithm is used to align the notes, and the Hamming distance between the hash values of each two corresponding notes is calculated to obtain the measurement values of error series, statistical characteristics, and time distribution characteristics. According to the measurement value, the fuzzy classification method is applied to calculate the membership degree of the signals belonging to two different types of operation, and the authentication confidence degree is obtained. The tampered area of the music signal that has not passed the authentication is detected, and the music multiterminal audio authentication is realized. Experimental results show that the proposed algorithm has good tamper detection ability and can effectively shorten the audio authentication time.

An Antiforensic Method against AMR Compression Detection

Adaptive multirate (AMR) compression audio has been exploited as an effective forensic evidence to justify audio authenticity. Little consideration has been given, however, to antiforensic techniques capable of fooling AMR compression forensic algorithms. In this paper, we present an antiforensic method based on generative adversarial network (GAN) to attack AMR compression detectors. The GAN framework is utilized to modify double AMR compressed audio to have the underlying statistics of single compressed one. Three state-of-the-art detectors of AMR compression are selected as the targets to be attacked. The experimental results demonstrate that the proposed method is capable of removing the forensically detectable artifacts of AMR compression under various ratios with an average successful attack rate about 94.75%, which means the modified audios generated by our well-trained generator can treat the forensic detector effectively. Moreover, we show that the perceptual quality of the generated AMR audio is well preserved.

Enhancing Learners’ Communicative Skills through Audio-Visual Means

Communication helps us to receive and transmit information with the help of the language. In the result of globalization and cultural integration English became a worldwide used language. Knowledge of English gives rise to numerous possibilities so modern specialists have to speak the English language fluently. Thus, methods of teaching students a foreign language should be developed. English language acquisition is closely connected with different types of activities with audio and video that help motivate students to communicate and enhance their communicative competence of English through listening and speaking. Moreover, listening to and watching authentic professionally-oriented educational materials has a better impact on the sensory organs and allows students to choose the appropriate vocabulary to respond quickly to communicative situations. Therefore, we hypothesize, that the effectiveness of the use of audiovisual means in teaching professional English communication depends on such factors as the teacher’s skills level, the quality of the audiovisual means used, the authenticity of the professional audio and video texts, the correct organization of the educational process and the consideration of the individual peculiarities of students. In this study we analyse theoretical data relevant to the current research, describe our personal approach to the choice of methods and techniques using audio and video to provide students with proficiency in professional English and seek to compare the level of English language proficiency before and after the experiment when conducting our lessons with different types of activities. Our findings prove that the appropriate usage of authentic audio and video professional materials by the teacher influences the sensory organs of students and allows them to memorize vocabulary and speech patterns better in order to react to communicative situations.

A source coding scheme for authenticating audio signal with capability of self-recovery and anti-synchronization counterfeiting attack

Authenticating the veracity and integrity of digital media content is the most important application of fragile watermarking technique. Recently, fragile watermarking schemes for digital audio signals are developed to not only detect the malicious falsification, but also recover the tampered audio content. However, they are fragile against synchronization counterfeiting attack, which greatly narrows the applicability of audio watermarking schemes. In this paper, a novel source coding scheme for authenticating audio signal based on set partitioning in hierarchical trees (SPIHT) encoding and chaotic dynamical system with capability of self-recovery and anti-synchronization counterfeiting attack is proposed. For self-recovery feature, the compressed version of audio signal generated by SPIHT source coding and protected against maliciously tampering by repeated coding is embedded into the original audio signal. Besides, for robustness against synchronization counterfeiting attack feature, based on the position and content of audio section, check bits are generated by Hash algorithm and chaotic sequence, and taken as part of fragile watermark. Simulation results show the self-embedding audio authentication scheme is recoverable with proper audio quality, and it has capability against synchronization counterfeiting attack.

AMR Compressed-Domain Analysis for Multimedia Forensics Double Compression Detection

An audio recording must be authentic to be admitted as evidence in a criminal prosecution so that the speech is saved with maximum fidelity and interpretation mistakes are prevented. AMR (adaptive multi-rate) encoder is a worldwide standard for speech compression and for GSM mobile network transmission, including 3G and 4G. In addition, such encoder is an audio file format standard with extension AMR which uses the same compression algorithm. Due to its extensive usage in mobile networks and high availability in modern smartphones, AMR format has been found in audio authenticity cases for forgery searching. Such exams compound the multimedia forensics field which consists of, among other techniques, double compression detection, i. e., to determine if a given AMR file was decompressed and compressed again. AMR double compression detection is a complex engineering problem whose solution is still underway. In general terms, if an AMR file is double compressed, it is not an original one and it was likely doctored. The published works in literature about double compression detection are based on decoded waveform AMR files to extract features. In this paper, a new approach is proposed to AMR double compression detection which, in spite of processing decoded audio, uses its encoded version to extract compressed-domain linear prediction (LP) coefficient-based features. By means of feature statistical analysis, it is possible to show that they can be used to achieve AMR double compression detection in an effective way, so that they can be considered a promising path to solve AMR double compression problem by artificial neural networks.

ДОСВІД ВИКОРИСТАННЯ SKYPE ЯК ЗАСОБУ ПІДВИЩЕННЯ ЕФЕКТИВНОСТІ НАВЧАННЯ ІНОЗЕМНИХ МОВ

The article deals with foreign languages teaching, namely German, with the help of the Skype communication system. In the XXI century, the use of interactive methods and Internet resources in the educative process is a promising direction for modern education modernization. The advantages of learning in Skype include the convenient schedule for classes both for the teacher and the student, personal freedom, residence interdependence, use of the full range of Internet resources, authenticity of texts, audio and video files, etc. Owning to the use of the full range of Internet resources, this kind of training improves monologic and dialogical speech, listening comprehension and writing skills, and promotes the development of various competencies. The most popular and useful sites containing materials for teaching German which facilitate and vary the work with the help of Skype are analyzed. Among them are Deutsche Welle; Der Weg. Portal für Deutschlernende; Deutschlernerblog für Deutschlerner in aller Welt; Mein Deutschbuch. The subject matter of these sites such as topics, grammar and lexical materials, audio and video are overviewed. The possibilities and advantages of each site are analyzed, for which purpose it is possible to use one or another section while learning with the help of Skype. Recommendations for using specific sites and sections to prepare for the German language exam to obtain an internationally recognized certificate are formulated. As an example, the article provides tutorials and links to specific sites for teaching in Skype on the topic “Food in our lives” at level A2. This lexical theme is proposed to be studied in parallel with such grammar materials as “Substantiv und Artikel”, “Perfekt”, “Präteritum”, and “Modalverben”. The author supplies links with grammar exercises that contain lexical material on the selected subject, which contributes to a better acquisition of new material and consolidating already learned grammatical themes. The article contains links to the Deutsche Welle website with lessons have been already developed within the lexical theme on-line and off-line, as well as topics for projects, tasks for socio-cultural development of students, proverbs and set expressions, links with texts for reading and listening comprehension.

An ENF-Based Audio Authenticity Method Robust to MP3 Compression

This work presents a novel method for assessing audio authenticity. Assuming that the electric network frequency is embedded in audio signals, the evaluation of audio integrity is carried out by detecting phase discontinuities. This is conducted by using causal and anti-causal filters, in order to avoid the mix of past and future phase information related to the time of analysis. This local phase change is then post-processed and thresholded to obtain the editing times. One remarkable property of the proposed method is its ability to withstand MP3 compression, an audio format widely used in practice. A more accurate evaluation metric is also introduced in this work. For that purpose, the databases used for evaluating the algorithm were automatically labeled indicating the editing times. The procedure to generate the ground truth is presented, as well as a discussion on the proposed metric. The performance of the technique presented promising results when evaluated on digitally edited and original audio signals.

Forensic audio authentication analysis technique of first or higher generation copies of analog magnetic audio tapes

In forensic audio examination, standard listening and visual sound spectrogram analysis methods often lead to incorrect conclusions due to the losses in data when the tape is copied. Furthermore, conducting the standard examinations on audio files which are digitally edited and re-recorded on the tapes, leaving very little trace behind, also result in incorrect decisions made by forensic experts. In this study, a novel and a very simple technique was developed, where the electronic edit traces on an electronically edited tape become easily detectable on copy generation tapes. Thus, regardless of the duplication speed, copied tapes can be analyzed for their originality and integrity with this technique and digital changes on them can be determined. The technique is based on increasing the tape playback speed by modifying the playback system of a player device. This essentially alters the time base of the signal and expands the low frequency scale of the recording in the frequency domain, and makes manipulation marks easily observable. It can be applied to most analog magnetic tape recorder devices. Edit marks can be precisely detected with minimal loss of time by using specially modified nonstandard devices rather than available conventional standard devices.

Authenticity verification of audio signals based on fragile watermarking for audio forensics

Abstract This paper presents a new fragile watermarking method for digital audio authenticity for audio forensics purposes. The aim is to verify if an audio proof has been tampered and to locate the segments where the signal was modified. Our proposal is based on an embedding process of a text that is encoded through OVSF (Orthogonal Variable Spreading Factor) codes and spread into the entire signal using automatic adjustment. Several tests were performed in order to quantify the accuracy and the reliability of the tampering detection against four classical attacks (cropping, replacement, additive noise and amplitude reduction) by using kappa index, sensitivity and specificity. It was demonstrated that even if a small number of samples is modified, the system correctly labels the audio proof as manipulated, and locates both the start and end of the manipulation; the kappa index (reliability) is around 0.96, sensitivity is always 1, and specificity is around 0.995. The proposed algorithm could be used as a decision support tool for audio forensics verification purposes, that allows to identify if an audio proof has been modified, and the time segments in which it has been modified.

ESPRIT-Hilbert-Based Audio Tampering Detection With SVM Classifier for Forensic Analysis via Electrical Network Frequency

Audio authentication is a critical task in multimedia forensics demanding robust methods to detect and identify tampered audio recordings. In this paper, a new technique to detect adulterations in audio recordings is proposed by exploiting abnormal variations in the electrical network frequency (ENF) signal eventually embedded in a questioned audio recording. These abnormal variations are caused by abrupt phase discontinuities due to insertions and suppressions of audio snippets during the tampering task. First, we propose an ESPRIT-Hilbert ENF estimator in conjunction with an outlier detector based on the sample kurtosis of the estimated ENF. Next, we use the computed kurtosis as an input for a support vector machine classifier to indicate the presence of tampering. The proposed scheme, herein designated as SPHINS, significantly outperforms related previous tampering detection approaches in the conducted tests. We validate our results using the Carioca 1 corpus with 100 unedited authorized audio recordings of phone calls.

An Automatic Digital Audio Authentication/Forensics System

With the continuous rise in ingenious forgery, a wide range of digital audio authentication applications are emerging as a preventive and detective control in real-world circumstances, such as forged evidence, breach of copyright protection, and unauthorized data access. To investigate and verify, this paper presents a novel automatic authentication system that differentiates between the forged and original audio. The design philosophy of the proposed system is primarily based on three psychoacoustic principles of hearing, which are implemented to simulate the human sound perception system. Moreover, the proposed system is able to classify between the audio of different environments recorded with the same microphone. To authenticate the audio and environment classification, the computed features based on the psychoacoustic principles of hearing are dangled to the Gaussian mixture model to make automatic decisions. It is worth mentioning that the proposed system authenticates an unknown speaker irrespective of the audio content i.e., independent of narrator and text. To evaluate the performance of the proposed system, audios in multi-environments are forged in such a way that a human cannot recognize them. Subjective evaluation by three human evaluators is performed to verify the quality of the generated forged audio. The proposed system provides a classification accuracy of 99.2% ± 2.6. Furthermore, the obtained accuracy for the other scenarios, such as text-dependent and text-independent audio authentication, is 100% by using the proposed system.

On Practical Issues of Electric Network Frequency Based Audio Forensics

The transmission frequency of power grids, i.e., electric network frequency (ENF), has become a common criterion to authenticate audio recordings during the past decade, drawing much attention from both the academic researchers and law enforcement agencies world widely. The properties of ENF enable forensic applications such as audio evidence timestamp verification and tampering detection. In this paper, based on a general review of existing works, we discuss several important practical problems and facts that have drawn less research attention or have not been formally studied, including ENF detection problems, limitations of the ENF-based tampering detection systems, and the difficulties in the ENF analysis. During ENF detection, the challenges come from not only the noise and the interference, but also the fact that audio recordings without captured ENF can still have signal components in the frequency band of interest (false positive). In ENF-based tampering detection systems, the weakness of commonly used assumptions and the limitations of several existing solutions are discussed. In addition, we reveal that in the most intensively studied ENF-based audio evidence timestamp verification, many works aiming at improving ENF estimation could only produce marginal performance improvement, while the main problems due to noise and interference remain open. All these analysis and discussions are related by a proposed big picture of ENF-based audio authentication systems. After that, we also investigate the strategies to design more reliable audio authentication systems based on the ENF, which consists of a series of research and investigation works.

Audio Authentication by Exploring the Absolute-Error-Map of ENF Signals

Recently, the electric network frequency (ENF), a natural signature embedded in many audio recordings, has been utilized as a criterion to examine the authenticity of audio recordings. ENF-based audio authentication system involves extraction of the ENF signal from a questioned audio recording, and matching it with the reference signal stored in an ENF database. This establishes a popular application of audio timestamp verification. In this paper, we explore another important application, i.e., ENF-based audio tampering detection, which has received less research attention. Specifically, we introduce the absolute-error-map (AEM) between the ENF signals obtained from the testing audio recording and the database. The AEM serves as an ensemble of the raw data associated with the ENF matching process. Through intensive analysis of the AEM, we propose two algorithms to jointly deal with timestamp verification and tampering detection, including insertion, deletion, and splicing attacks, respectively. The first algorithm is based on exhaustive point search and measurement, while the second algorithm leverages the image erosion technique to achieve fast detection of tampering type and tampered region, thus the second algorithm sacrifices some accuracy for speed. The authentication mechanism is that the system first determines if the testing data have been tampered with, and then outputs the timestamp information if no tampering is detected. Otherwise, it outputs the tampering type and tampered region. We demonstrate the effectiveness of the proposed solution via both synthetic and practical examples from our practically deployed audio authentication system.