- Showing results for:
- Reset all filters
Journal articleNeo VW, Redif S, McWhirter JG, et al., 2023,
This article is devoted to the polynomial eigenvalue decomposition (PEVD) and its applications in broadband multichannel signal processing, motivated by the optimum solutions provided by the eigenvalue decomposition (EVD) for the narrow-band case , . In general, the successful techniques from narrowband problems can also be applied to broadband ones, leading to improved solutions. Multichannel broadband signals arise at the core of many essential commercial applications such as telecommunications, speech processing, healthcare monitoring, astronomy and seismic surveillance, and military technologies like radar, sonar and communications . The success of these applications often depends on the performance of signal processing tasks, including data compression , source localization , channel coding , signal enhancement , beamforming , and source separation . In most cases and for narrowband signals, performing an EVD is the key to the signal processing algorithm. Therefore, this paper aims to introduce PEVD as a novel mathematical technique suitable for many broadband signal processing applications.
Journal articleNeo VW, Evers C, Weiss S, et al., 2023,
Multi-channel signals captured by spatially separated sensors often contain a high level of data redundancy. A compact signal representation enables more efficient storage and processing, which has been exploited for data compression, noise reduction, and speech and image coding. This paper focuses on the compact representation of speech signals acquired by spherical microphone arrays. A polynomial matrix eigenvalue decomposition (PEVD) can spatially decorrelate signals over a range of time lags and is known to achieve optimum multi-channel data compaction. However, the complexity of PEVD algorithms scales at best cubically with the number of channel signals, e.g., the number of microphones comprised in a spherical array used for processing. In contrast, the spherical harmonic transform (SHT) provides a compact spatial representation of the 3-dimensional sound field measured by spherical microphone arrays, referred to as eigenbeam signals, at a cost that rises only quadratically with the number of microphones. Yet, the SHT's spatially orthogonal basis functions cannot completely decorrelate sound field components over a range of time lags. In this work, we propose to exploit the compact representation offered by the SHT to reduce the number of channels used for subsequent PEVD processing. In the proposed framework for signal representation, we show that the diagonality factor improves by up to 7 dB over the microphone signal representation with a significantly lower computation cost. Moreover, when applying this framework to speech enhancement and source separation, the proposed method improves metrics known as short-time objective intelligibility (STOI) and source-to-distortion ratio (SDR) by up to 0.2 and 20 dB, respectively.
Journal articleGrinstein E, Neo VW, Naylor PA, 2023,
In many signal processing applications, metadata may be advantageously used in conjunction with a high dimensional signal to produce a desired output. In the case of classical Sound Source Localization (SSL) algorithms, information from a high dimensional, multichannel audio signals received by many distributed microphones is combined with information describing acoustic properties of the scene, such as the microphones’ coordinates in space, to estimate the position of a sound source. We introduce Dual Input Neural Networks (DI-NNs) as a simple and effective way to model these two data types in a neural network. We train and evaluate our proposed DI-NN on scenarios of varying difficulty and realism and compare it against an alternative architecture, a classical Least-Squares (LS) method as well as a classical Convolutional Recurrent Neural Network (CRNN). Our results show that the DI-NN significantly outperforms the baselines, achieving a five times lower localization error than the LS method and two times lower than the CRNN in a test dataset of real recordings.
Conference paperSanguedolce G, Naylor PA, Geranmayeh F, 2023,
Uncovering the potential for a weakly supervised end-to-end model in recognising speech from patient with post-stroke aphasia, 5th Clinical Natural Language Processing Workshop, Publisher: Association for Computational Linguistics, Pages: 182-190
Post-stroke speech and language deficits (aphasia) significantly impact patients' quality of life. Many with mild symptoms remain undiagnosed, and the majority do not receive the intensive doses of therapy recommended, due to healthcare costs and/or inadequate services. Automatic Speech Recognition (ASR) may help overcome these difficulties by improving diagnostic rates and providing feedback during tailored therapy. However, its performance is often unsatisfactory due to the high variability in speech errors and scarcity of training datasets. This study assessed the performance of Whisper, a recently released end-to-end model, in patients with post-stroke aphasia (PWA). We tuned its hyperparameters to achieve the lowest word error rate (WER) on aphasic speech. WER was significantly higher in PWA compared to age-matched controls (10.3% vs 38.5%, p < 0.001). We demonstrated that worse WER was related to the more severe aphasia as measured by expressive (overt naming, and spontaneous speech production) and receptive (written and spoken comprehension) language assessments. Stroke lesion size did not affect the performance of Whisper. Linear mixed models accounting for demographic factors, therapy duration, and time since stroke, confirmed worse Whisper performance with left hemispheric frontal lesions. We discuss the implications of these findings for how future ASR can be improved in PWA.
Conference paperNespoli F, Barreda D, Bitzer J, et al., 2023,
In recent years, the need for privacy preservation when manipulating or storing personal data, including speech, has become a major issue. In this paper, we present a system addressing the speaker-level anonymization problem. We propose and evaluate a two-stage anonymization pipeline exploiting a state-of-the-art anonymization model described in the Voice Privacy Challenge 2022 in combination with a zero-shot voice conversion architecture able to capture speaker characteristics from a few seconds of speech. We show this architecture can lead to strong privacy preservation while preserving pitch information. Finally, we propose a new compressed metric to evaluate anonymization systems in privacy scenarios with different constraints on privacy and utility.
Conference paperMcKnight S, Hogg AOT, Neo VW, et al., 2022,
Human-based speaker diarization experiments were carried out on a five-minute extract of a typical AMI corpus meeting to see how much variance there is in human reviews based on hearing only and to compare with state-of-the-art diarization systems on the same extract. There are three distinct experiments: (a) one with no prior information; (b) one with the ground truth speech activity detection (GT-SAD); and (c) one with the blank ground truth labels (GT-labels). The results show that most human reviews tend to be quite similar, albeit with some outliers, but the choice of GT-labels can make a dramatic difference to scored performance. Using the GT-SAD provides a big advantage and improves human review scores substantially, though small differences in the GT-SAD used can have a dramatic effect on results. The use of forgiveness collars is shown to be unhelpful. The results show that state-of-the-art systems can outperform the best human reviews when no prior information is provided. However, the best human reviews still outperform state-of-the-art systems when starting from the GT-SAD.
Conference paperD'Olne E, Neo VW, Naylor PA, 2022,
As the number of connected devices equipped withmultiple microphones increases, scientific interest in distributedmicrophone array processing grows. Current beamforming meth-ods heavily rely on estimating quantities related to array geom-etry, which is extremely challenging in real, non-stationary envi-ronments. Recent work on polynomial eigenvalue decomposition(PEVD) has shown promising results for speech enhancement insingular arrays without requiring the estimation of any array-related parameter . This work extends these results to therealm of distributed microphone arrays, and further presentsa novel framework for speech enhancement in distributed mi-crophone arrays using PEVD. The proposed approach is shownto almost always outperform optimum beamformers located atarrays closest to the desired speaker. Moreover, the proposedapproach exhibits very strong robustness to steering vectorerrors.
Conference paperNeo VW, Weiss S, McKnight S, et al., 2022,
Polynomial eigenvalue decomposition-based target speaker voice activity detection in the presence of competing talkers, International Workshop on Acoustic Signal Enhancement (IWAENC), Publisher: IEEE, Pages: 1-5
Voice activity detection (VAD) algorithms are essential for many speech processing applications, such as speaker diarization, automatic speech recognition, speech enhancement, and speech coding. With a good VAD algorithm, non-speech segments can be excluded to improve the performance and computation of these applications. In this paper, we propose a polynomial eigenvalue decomposition-based target-speaker VAD algorithm to detect unseen target speakers in the presence of competing talkers. The proposed approach uses frame-based processing to compute the syndrome energy, used for testing the presence or absence of a target speaker. The proposed approach is consistently among the best in F1 and balanced accuracy scores over the investigated range of signal to interference ratio (SIR) from -10 dB to 20 dB.
Conference paperNeo VW, D'Olne E, Moore AH, et al., 2022,
Array processing is widely used in many speech applications involving multiple microphones. These applications include automaticspeech recognition, robot audition, telecommunications, and hearing aids. A spatio-temporal filter for the array allows signals fromdifferent microphones to be combined desirably to improve the application performance. This paper will analyze and visually interpretthe eigenvector beamformers designed by the polynomial eigenvaluedecomposition (PEVD) algorithm, which are suited for arbitrary arrays. The proposed fixed PEVD beamformers are lightweight, withan average filter length of 114 and perform comparably to classicaldata-dependent minimum variance distortionless response (MVDR)and linearly constrained minimum variance (LCMV) beamformersfor the separation of sources closely spaced by 5 degrees.
Conference paperTokala V, Brookes M, Naylor P, 2022,
STOI-optimal masking has been previously proposed and developed for single-channel speech enhancement. In this paper, we consider the extension to the task of binaural speech enhancement in which the spatial information is known to be important to speech understanding and therefore should bepreserved by the enhancement processing. Masks are estimated for each of the binaural channels individually and a ‘better-ear listening’ mask is computed by choosing the maximum of the two masks. The estimated mask is used to supply probability information about the speech presence in eachtime-frequency bin to an Optimally-modified Log Spectral Amplitude (OM-LSA) enhancer. We show that using the pro-posed method for binaural signals with a directional noise not only improves the SNR of the noisy signal but also preserves the binaural cues and intelligibility.
Conference paperD'Olne E, Neo VW, Naylor PA, 2022,
Frame-based space-time covariance matrix estimation for polynomial eigenvalue decomposition-based speech enhancement, International Workshop on Acoustic Signal Enhancement (IWAENC), Publisher: IEEE, Pages: 1-5
Recent work in speech enhancement has proposed a polynomial eigenvalue decomposition (PEVD) method, yielding significant intelligibility and noise-reduction improvements without introducing distortions in the enhanced signal . The method relies on the estimation of a space-time covariance matrix, performed in batch mode such that a sufficiently long portion of the noisy signal is used to derive an accurate estimate. However, in applications where the scene is nonstationary, this approach is unable to adapt to changes in the acoustic scenario. This paper thus proposes a frame-based procedure for the estimation of space-time covariance matrices and investigates its impact on subsequent PEVD speech enhancement. The method is found to yield spatial filters and speech enhancement improvements comparable to the batch method in , showing potential for real-time processing.
Conference paperNeo VW, Weiss S, Naylor PA, 2022,
A voice activity detection (VAD) algorithm identifies whether or not time frames contain speech. It is essential for many military and commercial speech processing applications, including speech enhancement, speech coding, speaker identification, and automatic speech recognition. In this work, we adopt earlier work on detecting weak transient signals and propose a polynomial subspace projection pre-processor to improve an existing VAD algorithm. The proposed multi-channel pre-processor projects the microphone signals onto a lower dimensional subspace which attempts to remove the interferer components and thus eases the detection of the speech target. Compared to applying the same VAD to the microphone signal, the proposed approach almost always improves the F1 and balanced accuracy scores even in adverse environments, e.g. -30 dB SIR, which may be typical of operations involving noisy machinery and signal jamming scenarios.
Conference paperMcKnight S, Hogg A, Neo V, et al., 2022,
A study of salient modulation domain features for speaker identification, Asia Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), Publisher: IEEE, Pages: 705-712
This paper studies the ranges of acoustic andmodulation frequencies of speech most relevant for identifyingspeakers and compares the speaker-specific information presentin the temporal envelope against that present in the temporalfine structure. This study uses correlation and feature importancemeasures, random forest and convolutional neural network mod-els, and reconstructed speech signals with specific acoustic and/ormodulation frequencies removed to identify the salient points. Itis shown that the range of modulation frequencies associated withthe fundamental frequency is more important than the 1-16 Hzrange most commonly used in automatic speech recognition, andthat the 0 Hz modulation frequency band contains significantspeaker information. It is also shown that the temporal envelopeis more discriminative among speakers than the temporal finestructure, but that the temporal fine structure still contains usefuladditional information for speaker identification. This researchaims to provide a timely addition to the literature by identifyingspecific aspects of speech relevant for speaker identification thatcould be used to enhance the discriminant capabilities of machinelearning models.
Journal articleGreen T, Hilkhuysen G, Huckvale M, et al., 2022,
Speech recognition with a hearing-aid processing scheme combining beamforming with mask-informed speech enhancement, Trends in Hearing, Vol: 26, Pages: 1-16, ISSN: 2331-2165
A signal processing approach combining beamforming with mask-informed speech enhancement was assessed by measuring sentence recognition in listeners with mild-to-moderate hearing impairment in adverse listening conditions that simulated the output of behind-the-ear hearing aids in a noisy classroom. Two types of beamforming were compared: binaural, with the two microphones of each aid treated as a single array, and bilateral, where independent left and right beamformers were derived. Binaural beamforming produces a narrower beam, maximising improvement in signal-to-noise ratio (SNR), but eliminates the spatial diversity that is preserved in bilateral beamforming. Each beamformer type was optimised for the true target position and implemented with and without additional speech enhancement in which spectral features extracted from the beamformer output were passed to a deep neural network trained to identify time-frequency regions dominated by target speech. Additional conditions comprising binaural beamforming combined with speech enhancement implemented using Wiener filtering or modulation-domain Kalman filtering were tested in normally-hearing (NH) listeners. Both beamformer types gave substantial improvements relative to no processing, with significantly greater benefit for binaural beamforming. Performance with additional mask-informed enhancement was poorer than with beamforming alone, for both beamformer types and both listener groups. In NH listeners the addition of mask-informed enhancement produced significantly poorer performance than both other forms of enhancement, neither of which differed from the beamformer alone. In summary, the additional improvement in SNR provided by binaural beamforming appeared to outweigh loss of spatial information, while speech understanding was not further improved by the mask-informed enhancement method implemented here.
Conference paperSathyapriyan V, Pedersen MS, Ostergaard J, et al., 2022,
Conference paperJones DT, Sharma D, Kruchinin SY, et al., 2022,
Microphone Array Coding Preserving Spatial Information for Cloud-based Multichannel Speech Recognition, 30th European Signal Processing Conference (EUSIPCO), Publisher: IEEE, Pages: 324-328, ISSN: 2076-1465
Conference paperLi G, Sharma D, Naylor PA, 2022,
Non-Intrusive Signal Analysis for Room Adaptation of ASR Models, 30th European Signal Processing Conference (EUSIPCO), Publisher: IEEE, Pages: 130-134, ISSN: 2076-1465
Conference paperNespoli F, Barreda D, Naylor PA, 2022,
Conference paperGuiraud P, Moore AH, Vos RR, et al., 2022,
Conference paperSharma D, Gong R, Fosburgh J, et al., 2022,
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.