V
Vinayak Abrol
Researcher at University of Oxford
Publications - 49
Citations - 364
Vinayak Abrol is an academic researcher from University of Oxford. The author has contributed to research in topics: Sparse approximation & Speech processing. The author has an hindex of 10, co-authored 43 publications receiving 296 citations. Previous affiliations of Vinayak Abrol include Idiap Research Institute & University Institute of Engineering and Technology, Panjab University.
Papers
More filters
Journal ArticleDOI
Deep-Sparse-Representation-Based Features for Speech Recognition
TL;DR: This paper proposes to use a multilevel decomposition (having multiple layers), also known as the deep sparse representation (DSR), to derive a feature representation for speech recognition, and reveals that the representations obtained at different sparse layers of the proposed DSR model have complimentary information.
Proceedings ArticleDOI
Understanding and Visualizing Raw Waveform-based CNNs
TL;DR: This paper develops a gradient based approach to estimate the relevance of each speech sample input on the output score, and shows that analysis of the resulting “relevance signal” through conventional speech signal processing techniques can reveal the information modeled by the whole network.
Journal ArticleDOI
Voiced/nonvoiced detection in compressively sensed speech signals
TL;DR: The proposed novel unsupervised voiced/nonvoiced (V/NV) detection method attempts to exploit the fact that there is significant glottal activity during production of voiced speech while the same is not true for nonvoiced speech, and provides compelling evidence of the effectiveness of sparse feature vector for V/NV detection.
Journal ArticleDOI
Greedy dictionary learning for kernel sparse representation based classifier
TL;DR: Compared to the existing state-of-the-art methods, the proposed method has much less computational complexity, but performs similar for various pattern classification tasks.
Proceedings ArticleDOI
Evaluating performance of Compressed Sensing for speech signals
TL;DR: This work shows a comparative analysis of different sparse basis & measurement matrices which can be used in speech/audio processing and gives a detail analysis of the performance bounds, compression ratios, reconstruction errors etc. which should be taken care of while designing CS based speech applications.