LP residual features to counter replay attacks

Mishra, Jagabandhu; Singh, Madhusudan; Pati, Debadatta

doi:10.1109/icsigsys.2018.8372769

Cited by 8 publications

(7 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Proper choice of LP order eliminates the formant information from the residual spectrum and contains mostly excitation source information. If f s is the sampling frequency then the order of prediction p is given by f s +2 [25]. Therefore p is set to 18.…”

Section: A Lp Residualmentioning

confidence: 99%

See 1 more Smart Citation

Speaker Recognition in Emotional Environment using Excitation Features

Thomas¹,

Spoorthy

Sobhana³

et al. 2020

2020 Third International Conference on Advances in Electronics, Computers and Communications (ICAECC)

View full text Add to dashboard Cite

Speaker Recognition is known as the task of recognizing the person speaking from his/her speech. Speaker recognition has many applications including transaction authentication, access control, voice dialing, web services, etc. Emotive speaker recognition is important because in real life, human beings extensively express emotions during conversations, and emotions alter the human voice. A text-independent speaker recognition system is proposed in the work. The system designed is for emotional environment. The proposed system in this work is trained using the speech samples recorded in neutral environment and the system evaluation is performed in an emotional environment. Here, excitation source features are used to represent speakerspecific details contained in speech signal. The excitation source signal is obtained after separating the segmental level features from the voice samples. The excitation source signal is almost considered as a noise so identifying a speaker in an emotive environment is a challenging task. Excitation features include Linear Prediction (LP) residual, Glottal Closure Instance (GCI), LP residual phase, residual cepstrum, Residual Mel-Frequency Cepstral Coefficient (R-MFCC), etc. A decrease in performance is observed when the system is trained with neutral speech samples and tested with emotional speech samples. Different emotions considered for emotional speaker identification are happy, sad, anger, fear, neutral, surprise, disgust, and sarcastic For the classification of speakers the algorithms used are Gaussian Mixture Model (GMM), Support Vector Machine (SVM), K-Nearest Neighbor(KNN), Random Forest and Naive Bayes.

show abstract

Section: A Lp Residualmentioning

confidence: 99%

“…MFCC contain vocal tract information and R-MFCC contain excitation source information about the speaker. To compute R-MFCC first, the log magnitude spectrum of residual signal is computed and the resultant is wrapped with mel filter [25]. Then the cepstral coefficients are obtained by applying inverse discrete fourier transform.…”

Section: E R-mfccmentioning

confidence: 99%

Speaker Recognition in Emotional Environment using Excitation Features

Thomas¹,

Spoorthy

Sobhana³

et al. 2020

2020 Third International Conference on Advances in Electronics, Computers and Communications (ICAECC)

View full text Add to dashboard Cite

show abstract

“…In recent studies [21], [25], [27], the magnitude of LPR spectra via the discrete Fourier transform (DFT) has been proven to be powerful for replay attack detection. This is because the magnitude of replayed LPR spectra is highly affected by the distortion of recording and playback devices, which is different from the magnitude of genuine LPR spectra.…”

Section: The Motivation For Using the Phase Information Based On A Linear Prediction Analysis-based Speech Signalmentioning

confidence: 99%

“…In all these features, the decision for detecting genuine speech from replay speech is strongly based on the magnitude spectrum information of short-time spectral analysis obtained from original/raw speech. However, the magnitude spectrum information of original/raw speech via short-time spectral analysis is weak for detecting replayed signals as modern recording, and playback devices can maintain the average spectral patterns of original speech in replayed signals [21]. Thus, a source of alternative information for feature extraction is required to counter the replayed signal.…”

Section: Introductionmentioning

confidence: 99%

Replay Attack Detection Using Linear Prediction Analysis-Based Relative Phase Features

et al. 2019

View full text Add to dashboard Cite

Recent studies have reported the success of linear prediction analysis (LPA)-related features, which are extracted as a short-term spectral feature for replay attack detection due to the advantage of the imperfection in the LPA-based signal produced by recording and playback devices. However, exploiting LPA-based signals is focused on only magnitude-based features and ignores phase-based features. In this paper, we propose two novel LPA-based relative phase features, namely, linear prediction residual-based relative phase (LPR-RP) and linear prediction analysis estimated speech-based relative phase (LPAES-RP).The key idea of both LPR-RP and LPAES-RP is to extract the phase information based on LPA-based signals. In the LPR-RP feature, we modify the relative phase (RP) feature extraction using a linear prediction residual (LPR) derived from the difference between the original/raw speech and LPA estimated speech signal (LPAES) instead of the original/raw speech signal. LPES-RP feature exploits the LPAES signal to replace the original/raw speech signal. Because the trace of the recording and playback device artifacts is the primary evidence for detecting the replayed signal, the advantages of the imperfection of LPR and LPAES are expected to provide efficient phase information for the replay attack detection task. In addition, using the individual LPR-RP/LPAES-RP feature, our proposed features are combined with two standard features, mel-frequency cepstral coefficients (MFCC), constant Q transform cepstral coefficients (CQCC) and the original RP feature, at score level to further improve the detection decision. The performance of the proposed LPR-RP/LPAES-RP feature and combination are evaluated using the ASVspoof 2017 version 2 database. On the evaluation subset, our proposed LPR-RP and LPAES-RP feature achieves a promising improvement over baseline features (MFCC/CQCC). Moreover, the combined systems of LPR-RP, RP, and CQCC obtains an equal error rate of 9.26%. INDEX TERMSReplay attack detection, phase information, speaker recognition anti-spoofing, linear prediction analysis-based feature.

show abstract

“…In Yu, Tan, Ma, Martin, and Guo (2017) authors use different types of filter banks such as linear, Gammatone, and its inverted version and also inverted Mel filter banks to extract different variations of cepstral coefficients for replay attack detection. Furthermore, recently, linear prediction residual-based features like linear prediction residual magnitude cepstral coefficients (Hanilç, 2018), linear prediction residual phase cepstral coefficients (Hanilç, 2018), LP-based relative phase features (Phapatanaburi, Wang, Nakagawa, & Iwahashi, 2019), and residual Mel frequency cepstral coefficients (Mishra, Singh, & Pati, 2018), have shown interesting and promising results in detection of spoofed speech.…”

mentioning

confidence: 99%

Evolutionary fusion of classifiers trained on linear prediction based features for replay attack detection

Nasersharif

Yazdani

2021

Expert Systems

View full text Add to dashboard Cite

Recently, linear prediction analysis (LP) related features have been successfully used for replay attack detection due to the imperfection in the LP‐based signal produced by recording and playback devices. In this paper, we propose a weighted linear combination of classifier scores for replay attack detection where our classifiers, including Gaussian mixture models (GMMs) and support vector machines (SVMs), are trained on a variety of LP and LP residual‐based features. In this way, we can benefits from all of the LP‐related features when we combine classifiers trained on these features. We determine classifier weights using two evolutionary algorithms: genetic algorithm and particle swarm optimization. Furthermore, we propose a new feature based on performing LP residuals analysis of Mel sub‐band energies. We also propose a deep structure for extracting deep features from LP‐based coefficients to consider the class labels (genuine or spoofed speaker) in the feature extraction process. Results of our classifier system on the ASVspoof 2017 version 2 dataset show equal error rates of 0.3% and 4.8% for its development and evaluation subset, respectively. We also applied our proposed replay attack detection method to an ASV system that has acceptable results.

show abstract

LP residual features to counter replay attacks

Cited by 8 publications

References 11 publications

Speaker Recognition in Emotional Environment using Excitation Features

Speaker Recognition in Emotional Environment using Excitation Features

Replay Attack Detection Using Linear Prediction Analysis-Based Relative Phase Features

Evolutionary fusion of classifiers trained on linear prediction based features for replay attack detection

Contact Info

Product

Resources

About