APAmir Hossein Poorjam

Amir Hossein Poorjam

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Amir Hossein Poorjam has a total of 23 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Gradient-Free Post-Hoc Explainability Using Distillation Aided Learnable Approach

    by Debarpan Bhattacharya, Amir H. Poorjam, Deepak Mittal, Sriram Ganapathy

    The recent advancements in artificial intelligence (AI), with the release of several large models having only query access, make a strong case for explainability of deep models in a post-hoc gradient free manner. In this paper, we propose a framework, named distillation aided explainability (DAX), that attempts to generate a saliency-based explanation in a model agnostic gradient free application. The DAX approach poses the problem of explanation in a learnable setting with a mask generation network and a distillation network. The mask generation network learns to generate the multiplier mask that finds the salient regions of the input, while the student distillation network aims to approximate the local behavior of the black-box model. We propose a joint optimization of the two networks in the DAX framework using the locally perturbed input samples, with the targets derived from input-output access to the black-box model. We extensively evaluate DAX across different modalities (image and audio), in a classification setting, using a diverse set of evaluations (intersection over union with ground truth, deletion based and subjective human evaluation based measures) and benchmark it with respect to 9 different methods. In these evaluations, the DAX significantly outperforms the existing approaches on all modalities and evaluation metrics.


  • Open Access

    Interpretable Acoustic Representation Learning on Breathing and Speech Signals for COVID-19 Detection

    by Debottam Dutta, Debarpan Bhattacharya, Sriram Ganapathy, Amir H. Poorjam, Deepak Mittal, Maneesh Singh

    In this paper, we describe an approach for representation learning of audio signals for the task of COVID-19 detection. The raw audio samples are processed with a bank of 1-D convolutional filters that are parameterized as cosine modulated Gaussian functions. The choice of these kernels allows the interpretation of the filterbanks as smooth band-pass filters. The filtered outputs are pooled, log-compressed and used in a self-attention based relevance weighting mechanism. The relevance weighting emphasizes the key regions of the time-frequency decomposition that are important for the downstream task. The subsequent layers of the model consist of a recurrent architecture and the models are trained for a COVID-19 detection task. In our experiments on the Coswara data set, we show that the proposed model achieves significant performance improvements over the baseline system as well as other representation learning approaches. Further, the approach proposed is shown to be uniformly applicable for speech and breathing signals and for transfer learning from a larger data set.

  • Automatic quality control and enhancement for voice-based remote Parkinson’s disease detection

    by Amir Hossein Poorjam, Mathew Shaji Kavalekalam, Liming Shi, Jordan P. Raykov, Jesper Rindom Jensen, Max A. Little, Mads Græsbøll Christensen


  • Open Access

    A CNN-based approach to identification of degradations in speech signals

    by Yuki Saishu, Amir Hossein Poorjam, Mads Græsbøll Christensen

    The presence of degradations in speech signals, which causes acoustic mismatch between training and operating conditions, deteriorates the performance of many speech-based systems. A variety of enhancement techniques have been developed to compensate the acoustic mismatch in speech-based applications. To apply these signal enhancement techniques, however, it is necessary to know prior information about the presence and the type of degradations in speech signals. In this paper, we propose a new convolutional neural network (CNN)-based approach to automatically identify the major types of degradations commonly encountered in speech-based applications, namely additive noise, nonlinear distortion, and reverberation. In this approach, a set of parallel CNNs, each detecting a certain degradation type, is applied to the log-mel spectrogram of audio signals. Experimental results using two different speech types, namely pathological voice and normal running speech, show the effectiveness of the proposed method in detecting the presence and the type of degradations in speech signals which outperforms the state-of-the-art method. Using the score weighted class activation mapping, we provide a visual analysis of how the network makes decision for identifying different types of degradation in speech signals by highlighting the regions of the log-mel spectrogram which are more influential to the target degradation.


  • Quality Control of Voice Recordings in Remote Parkinson’s Disease Monitoring Using the Infinite Hidden Markov Model

    by Amir Hossein Poorjam, Yordan P. Raykov, Reham Badawy, Jesper Rindom Jensen, Mads Grasboll Christensen, Max A. Little

    The performance of voice-based systems for remote monitoring of Parkinson's disease is highly dependent on the degree of adherence of the recordings to the test protocols, which probe for specific symptoms. Identifying segments of the signal that adhere to the protocol assumptions is typically performed manually by experts. This process is costly, time consuming, and often infeasible for large-scale data sets. In this paper, we propose a method to automatically identify the segments of signals that violate the test protocol with a high accuracy. In our approach, the signal is first split into variable duration segments by fitting an infinite hidden Markov model (iHMM) to the frames of the signals in the mel-frequency cepstral domain. The complexity of the iHMM is capable of growing jointly with the data allowing us to infer a potentially large (asymptotically infinite) number of different phenomena segmented into different hidden states. Then, we identify the segments that adhere to the test protocol by applying a multinomial naive Bayes classifier to the state indicators of segments. The experimental results show that even by using a small amount of training data, we can achieve around 96% accuracy in identifying short-term protocol violations with a 0.2 s resolution.


  • Quality Control in Remote Speech Data Collection

    by Amir Hossein Poorjam, Max A. Little, Jesper Rindom Jensen, Mads Graesboll Christensen

    There is the need for algorithms that can automatically control the quality of the remotely collected speech databases by detecting potential outliers which deserve further investigation. In this paper, a simple and effective approach for identification of outliers in a speech database is proposed. Using the deterministic minimum covariance determinant (DetMCD) algorithm to estimate the mean and covariance of the speech data in the mel-frequency cepstral domain, this approach identifies potential outliers based on the statistical distance of the observations in the feature space from the central location of the data that are larger than a predefined threshold. The DetMCD is a computationally efficient algorithm which provides a highly robust estimate of the mean and covariance in multivariate data even when 50% of the data are outliers. Experimental results using 8 different speech databases with manually inserted outliers show the effectiveness of the proposed method for outlier detection in speech databases. Moreover, applying the proposed method to a remotely collected Parkinson's voice database shows that the outliers that are part of the database are detected with 97.4% accuracy, resulting in significantly decreasing the effort required for manually controlling the quality of the database.


  • A Supervised Approach to Global Signal-to-Noise Ratio Estimation for Whispered and Pathological Voices

    by Amir Hossein Poorjam, Max A. Little, Jesper Rindom Jensen, Mads Graesboll Christensen

    The presence of background noise in signals adversely affects the performance of many speech-based algorithms. Accurate estimation of signal-to-noise-ratio (SNR), as a measure of noise level in a signal, can help in compensating for noise effects. Most existing SNR estimation methods have been developed for normal speech and might not provide accurate estimation for special speech types such as whispered or disordered voices, particularly, when they are corrupted by non-stationary noises. In this paper, we first investigate the impact of stationary and non-stationary noise on the behavior of mel-frequency cepstral coefficients (MFCCs) extracted from normal, whispered and pathological voices. We demonstrate that, regardless of the speech type, the mean and the covariance of MFCCs are predictably modified by additive noise and the amount of change is related to the noise level. Then, we propose a new supervised method for SNR estimation which is based on a regression model trained on MFCCs of the noisy signals. Experimental results show that the proposed approach provides accurate estimation and consistent performance for various speech types under different noise conditions.


  • A Parametric Approach for Classification of Distortions in Pathological Voices

    by Amir Hossein Poorjam, Max A. Little, Jesper Rindom Jensen, Mads Graesboll Christensen

    In biomedical acoustics, distortion in voice signals, commonly present during acquisition and transmission, adversely affects acoustic features extracted from pathological voice. Information on the type of distortion can help in compensating for its effects. This paper proposes a new approach to detecting four major types of commonly encountered distortion in remote analysis of pathological voice, namely background noise, reverberation, clipping and coding. In this approach, by applying factor analysis to Gaussian mixture model mean supervectors, distortions in variable-duration recordings are modeled by fixed-length, low-dimensional channel vectors. Then, linear discriminant analysis (LDA) is used to remove the remaining nuisance effects in the channel vectors. Finally, two different classifiers, namely support vector machines and probabilistic LDA classify the different types of distortion. Experimental results obtained using Parkinson's voices, as an example of pathological voice, show 11.4% relative improvement in performance over systems which directly use acoustic features for distortion classification.