Erfan Loweimi
Research Subjects
Erfan Loweimi has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Zhengjun Yue
Shenzhen Loop Area Institute, Shenzhen, China
4 shared publications - Zoran Cvetkovic
Department of Engineering, Kings College London, London, U.K.
4 shared publications - Peter Bell
School of Informatics, University of Edinburgh, Edinburgh, United Kingdom
3 shared publications - Steve Renals
Centre for Speech Technology Research (CSTR), University of Edinburgh, Edinburgh, U.K.
3 shared publications - Jon Barker
University of Sheffield
2 shared publications - Thomas Hain
The University of Sheffield
2 shared publications
Latest Publications
- 23 October 2026
- Published by Elsevier in Computer Speech & Language
- Vol. 103
Deep Scattering Spectrum for dysarthric speech recognition, detection and classification
- 23 October 2026
- Published by Elsevier in Computer Speech & Language
- Vol. 103
Deep Scattering Spectrum for dysarthric speech recognition, detection and classification
- Open Access
- 19 March 2026
- Published by Elsevier in SSRN Electronic Journal
Deep Scattering Spectrum for Dysarthric Speech Recognition, Detection and Classification
This paper addresses the tasks of dysarthric speech recognition, detection, and severity classification by leveraging Deep Scattering Spectrum (DSS) features. Our study explores various DSS representations with diverse time and frequency resolutions, as well as different wavelet configurations, to optimise speech representation. We analyse the performance of DSS features using the word error rates of acoustic models within both a hybrid ASR architecture, comprising a cascade of convolutional, bidirectional recurrent, and fully connected layers, and an end-to-end Conformer-based acoustic model, all trained from scratch using in-domain data only.Experiments are conducted on the widely-used TORGO and UASpeech databases, with analyses stratified by dysarthria severity levels (severe, moderate, mild, and typical speech). Additionally, we perform confusion matrix analysis and layer-wise probing %on the acoustic model trained on the speech recognition task, in the detection and classification applications to gain deeper insights into how dysarthria-related information is encoded along the acoustic model pipeline. Our findings demonstrate the effectiveness of the DSS features over a filterbank-based (FBank) baseline, achieving WERs of 30.0\% for dysarthric speech and 8.4\% for typical speech on the TORGO dataset, corresponding to relative WER reductions of 12.5\% and 21.5\%, respectively, and 30.2\% on UASpeech, yielding a 15.2\% relative reduction.
- 19 March 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessDeep Scattering Spectrum for Dysarthric Speech Recognition, Detection and Classification
This paper addresses the tasks of dysarthric speech recognition, detection, and severity classification by leveraging Deep Scattering Spectrum (DSS) features. Our study explores various DSS representations with diverse time and frequency resolutions, as well as different wavelet configurations, to optimise speech representation. We analyse the performance of DSS features using the word error rates of acoustic models within both a hybrid ASR architecture, comprising a cascade of convolutional, bidirectional recurrent, and fully connected layers, and an end-to-end Conformer-based acoustic model, all trained from scratch using in-domain data only.Experiments are conducted on the widely-used TORGO and UASpeech databases, with analyses stratified by dysarthria severity levels (severe, moderate, mild, and typical speech). Additionally, we perform confusion matrix analysis and layer-wise probing %on the acoustic model trained on the speech recognition task, in the detection and classification applications to gain deeper insights into how dysarthria-related information is encoded along the acoustic model pipeline. Our findings demonstrate the effectiveness of the DSS features over a filterbank-based (FBank) baseline, achieving WERs of 30.0\% for dysarthric speech and 8.4\% for typical speech on the TORGO dataset, corresponding to relative WER reductions of 12.5\% and 21.5\%, respectively, and 30.2\% on UASpeech, yielding a 15.2\% relative reduction.
- 26 November 2025
- Published by Elsevier in Flow Measurement and Instrumentation
- Vol. 108
Transfer learning for data-driven wet gas flow metering: Enhancing generalisation in digital measurement systems
- 26 November 2025
- Published by Elsevier in Flow Measurement and Instrumentation
- Vol. 108
Transfer learning for data-driven wet gas flow metering: Enhancing generalisation in digital measurement systems
- 8 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 31 (23299290) , 3346-3361
Phonetic Error Analysis Beyond Phone Error Rate
In this paper, we analyse the performance of the TIMIT-based phone recognition systems beyond the overall phone error rate (PER) metric. We consider three broad phonetic classes (BPCs): {affricate, diphthong, fricative, nasal, plosive, semi-vowel, vowel, silence}, {consonant, vowel, silence} and {voiced, unvoiced, silence} and, calculate the contribution of each phonetic class in terms of the substitution, deletion, insertion and PER. Furthermore, for each BPC we investigate the following: evolution of PER during training, effect of noise (NTIMIT), importance of different spectral subbands (1, 2, 4, and 8 kHz), usefulness of bidirectional vs unidirectional sequential modelling, transfer learning from WSJ and regularisation via monophones. In addition, we construct a confusion matrix for each BPC and analyse the confusions via dimensionality reduction to 2D at the input (acoustic features) and output (logits) levels of the acoustic model. We also compare the performance and confusion matrices of the BLSTM-based hybrid baseline system with those of the GMM-HMM based hybrid, Conformer and wav2vec 2.0 based end-to-end phone recognisers. Finally, the relationship of the unweighted and weighted PERs with the broad phonetic class priors is studied for both the hybrid and end-to-end systems.
- 8 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 31 (23299290) , 3346-3361
Phonetic Error Analysis Beyond Phone Error Rate
In this paper, we analyse the performance of the TIMIT-based phone recognition systems beyond the overall phone error rate (PER) metric. We consider three broad phonetic classes (BPCs): {affricate, diphthong, fricative, nasal, plosive, semi-vowel, vowel, silence}, {consonant, vowel, silence} and {voiced, unvoiced, silence} and, calculate the contribution of each phonetic class in terms of the substitution, deletion, insertion and PER. Furthermore, for each BPC we investigate the following: evolution of PER during training, effect of noise (NTIMIT), importance of different spectral subbands (1, 2, 4, and 8 kHz), usefulness of bidirectional vs unidirectional sequential modelling, transfer learning from WSJ and regularisation via monophones. In addition, we construct a confusion matrix for each BPC and analyse the confusions via dimensionality reduction to 2D at the input (acoustic features) and output (logits) levels of the acoustic model. We also compare the performance and confusion matrices of the BLSTM-based hybrid baseline system with those of the GMM-HMM based hybrid, Conformer and wav2vec 2.0 based end-to-end phone recognisers. Finally, the relationship of the unweighted and weighted PERs with the broad phonetic class priors is studied for both the hybrid and end-to-end systems.
- 26 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 31 (23299290) , 876-890
Multi-Stream Acoustic Modelling Using Raw Real and Imaginary Parts of the Fourier Transform
In this paper, we investigate multi-stream acoustic modelling using the raw real and imaginary parts of the Fourier transform of speech signals. Using the raw magnitude spectrum, or features derived from it, as a proxy for the real and imaginary parts leads to irreversible information loss and suboptimal information fusion. We discuss and quantify the importance of such information in terms of speech quality and intelligibility. In the proposed framework, the real and imaginary parts are treated as two streams of information, pre-processed via separate convolutional networks, and then combined at an optimal level of abstraction, followed by further post-processing via recurrent and fully-connected layers. The optimal level of information fusion in various architectures, training dynamics in terms of cross-entropy loss, frame classification accuracy and WER as well as the shape and properties of the filters learned in the first convolutional layer of single- and multi-stream models are analysed. We investigated the effectiveness of the proposed systems in various tasks: TIMIT/NTIMIT (phone recognition), Aurora-4 (noise robustness), WSJ (read speech), AMI (meeting) and TORGO (dysarthric speech). Across all tasks we achieved competitive performance: in Aurora-4, down to 4.6% WER on average, in WSJ down to 4.6% and 6.2% WERs for Eval-92 and Eval-93, for Dev/Eval sets of the AMI-IHM down to 23.3%/23.8% WERs and in the AMI-SDM down to 43.7%/47.6% WERs have been achieved. In TORGO, for dysarthric and typical speech we achieved down to 31.7% and 10.2% WERs, respectively.
- 26 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 31 (23299290) , 876-890
Multi-Stream Acoustic Modelling Using Raw Real and Imaginary Parts of the Fourier Transform
In this paper, we investigate multi-stream acoustic modelling using the raw real and imaginary parts of the Fourier transform of speech signals. Using the raw magnitude spectrum, or features derived from it, as a proxy for the real and imaginary parts leads to irreversible information loss and suboptimal information fusion. We discuss and quantify the importance of such information in terms of speech quality and intelligibility. In the proposed framework, the real and imaginary parts are treated as two streams of information, pre-processed via separate convolutional networks, and then combined at an optimal level of abstraction, followed by further post-processing via recurrent and fully-connected layers. The optimal level of information fusion in various architectures, training dynamics in terms of cross-entropy loss, frame classification accuracy and WER as well as the shape and properties of the filters learned in the first convolutional layer of single- and multi-stream models are analysed. We investigated the effectiveness of the proposed systems in various tasks: TIMIT/NTIMIT (phone recognition), Aurora-4 (noise robustness), WSJ (read speech), AMI (meeting) and TORGO (dysarthric speech). Across all tasks we achieved competitive performance: in Aurora-4, down to 4.6% WER on average, in WSJ down to 4.6% and 6.2% WERs for Eval-92 and Eval-93, for Dev/Eval sets of the AMI-IHM down to 23.3%/23.8% WERs and in the AMI-SDM down to 43.7%/47.6% WERs have been achieved. In TORGO, for dysarthric and typical speech we achieved down to 31.7% and 10.2% WERs, respectively.
- Open Access
- 23 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 30 (23299290) , 2968-2980
Acoustic Modelling From Raw Source and Filter Components for Dysarthric Speech Recognition
Acoustic modelling for automatic dysarthric speech recognition (ADSR) is a challenging task. Data deficiency is a major problem and substantial differences between typical and dysarthric speech complicate the transfer learning. In this paper, we aim at building acoustic models using the raw magnitude spectra of the source and filter components for ADSR. The proposed multi-stream models consist of convolutional, recurrent and fully-connected layers allowing for pre-processing various information streams and fusing them at an optimal level of abstraction. We demonstrate that such a multi-stream processing leverages information encoded in the vocal tract and excitation components and leads to normalising nuisance factors such as speaker attributes and speaking style. This leads to a better handling of dysarthric speech that exhibits large inter- and intra-speaker variabilities and results in a notable performance gain. Furthermore, we analyse the learned convolutional filters and visualise the outputs of different layers after dimensionality reduction to demonstrate how the speaker-related attributes are normalised along the pipeline. We also compare the proposed multi-stream model with various systems based on MFCC, FBank, raw waveform and i-vector, and, study the training dynamics as well as usefulness of the feature normalisation and data augmentation via speed perturbation. On the widely used TORGO and UASpeech dysarthric speech corpora, the proposed approach leads to a competitive performance of up to 35.3% and 30.3% WERs for dysarthric speech, respectively.
- 23 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 30 (23299290) , 2968-2980
Open AccessAcoustic Modelling From Raw Source and Filter Components for Dysarthric Speech Recognition
Acoustic modelling for automatic dysarthric speech recognition (ADSR) is a challenging task. Data deficiency is a major problem and substantial differences between typical and dysarthric speech complicate the transfer learning. In this paper, we aim at building acoustic models using the raw magnitude spectra of the source and filter components for ADSR. The proposed multi-stream models consist of convolutional, recurrent and fully-connected layers allowing for pre-processing various information streams and fusing them at an optimal level of abstraction. We demonstrate that such a multi-stream processing leverages information encoded in the vocal tract and excitation components and leads to normalising nuisance factors such as speaker attributes and speaking style. This leads to a better handling of dysarthric speech that exhibits large inter- and intra-speaker variabilities and results in a notable performance gain. Furthermore, we analyse the learned convolutional filters and visualise the outputs of different layers after dimensionality reduction to demonstrate how the speaker-related attributes are normalised along the pipeline. We also compare the proposed multi-stream model with various systems based on MFCC, FBank, raw waveform and i-vector, and, study the training dynamics as well as usefulness of the feature normalisation and data augmentation via speed perturbation. On the widely used TORGO and UASpeech dysarthric speech corpora, the proposed approach leads to a competitive performance of up to 35.3% and 30.3% WERs for dysarthric speech, respectively.
- Open Access
- 6 June 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2750-2754
Train Your Classifier First: Cascade Neural Networks Training from Upper Layers to Lower Layers
Although the lower layers of a deep neural network learn features which are transferable across datasets, these layers are not transferable within the same dataset. That is, in general, freezing the trained feature extractor (the lower layers) and retraining the classifier (the upper layers) on the same dataset leads to worse performance. In this paper, for the first time, we show that the frozen classifier is transferable within the same dataset. We develop a novel top-down training method which can be viewed as an algorithm for searching for high-quality classifiers. We tested this method on automatic speech recognition (ASR) tasks and language modelling tasks. The proposed method consistently improves recurrent neural network ASR models on Wall Street Journal, self-attention ASR models on Switchboard, and AWD-LSTM language models on WikiText-2.
- 6 June 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2750-2754
Open AccessTrain Your Classifier First: Cascade Neural Networks Training from Upper Layers to Lower Layers
Although the lower layers of a deep neural network learn features which are transferable across datasets, these layers are not transferable within the same dataset. That is, in general, freezing the trained feature extractor (the lower layers) and retraining the classifier (the upper layers) on the same dataset leads to worse performance. In this paper, for the first time, we show that the frozen classifier is transferable within the same dataset. We develop a novel top-down training method which can be viewed as an algorithm for searching for high-quality classifiers. We tested this method on automatic speech recognition (ASR) tasks and language modelling tasks. The proposed method consistently improves recurrent neural network ASR models on Wall Street Journal, self-attention ASR models on Switchboard, and AWD-LSTM language models on WikiText-2.
- Open Access
- 1 April 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 4824-4828
Exploring the Use of Group Delay for Generalised VTS Based Noise Compensation
In earlier work we studied the effect of statistical normalisation for phase-based features and observed it leads to a significant robustness improvement. This paper explores the extension of the generalised Vector Taylor Series (gVTS) noise compensation approach to the group delay (GD) domain. We discuss the problems it presents, propose some solutions and derive the corresponding formulae. Furthermore, the effects of additive and channel noise in the GD domain were studied. It was observed that the GD of the noisy observation is a convex combination of the GDs of the clean signal and the additive noise and also in the expected sense, channel GD tends to zero. Experiments on Aurora-4 showed that, despite training only on the clean speech, the proposed features provide average WER reductions of 0.8% absolute and 4.1% relative compared to an MFCC-based system trained on the multi-style data. Combining the gVTS with a bottleneck DNN-based system led to average absolute (relative) WER improvements of 6.0% (23.5%) when training on clean data and 2.5% (13.8%) when using multi-style training with additive noise.
- 1 April 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 4824-4828
Open AccessExploring the Use of Group Delay for Generalised VTS Based Noise Compensation
In earlier work we studied the effect of statistical normalisation for phase-based features and observed it leads to a significant robustness improvement. This paper explores the extension of the generalised Vector Taylor Series (gVTS) noise compensation approach to the group delay (GD) domain. We discuss the problems it presents, propose some solutions and derive the corresponding formulae. Furthermore, the effects of additive and channel noise in the GD domain were studied. It was observed that the GD of the noisy observation is a convex combination of the GDs of the clean signal and the additive noise and also in the expected sense, channel GD tends to zero. Experiments on Aurora-4 showed that, despite training only on the clean speech, the proposed features provide average WER reductions of 0.8% absolute and 4.1% relative compared to an MFCC-based system trained on the multi-style data. Combining the gVTS with a bottleneck DNN-based system led to average absolute (relative) WER improvements of 6.0% (23.5%) when training on clean data and 2.5% (13.8%) when using multi-style training with additive noise.
- Open Access
- 11 February 2015
- Published by University of Sheffield Conference Proceedings
A Novel Phase-based Feature for Robust Speech Recognition
- 11 February 2015
- Published by University of Sheffield Conference Proceedings
Open AccessA Novel Phase-based Feature for Robust Speech Recognition