Results
- Showing results for:
- Reset all filters
Search results
-
Journal articleHogg A, Jenkins M, Liu H, et al., 2024, , IEEE Transactions on Audio, Speech and Language Processing, Vol: 32, Pages: 2085-2099, ISSN: 1558-7916
An individualised (HRTF) is very important for creating realistic (VR) and (AR) environments. However, acoustically measuring high-quality HRTFs requires expensive equipment and an acoustic lab setting. To overcome these limitations and to make this measurement more efficient HRTF upsampling has been exploited in the past where a high-resolution HRTF is created from a low-resolution one. This paper demonstrates how (GAN) can be applied to HRTF upsampling. We propose a novel approach that transforms the HRTF data for direct use with a convolutional (SRGAN). This new approach is benchmarked against three baselines: barycentric upsampling, (SH) upsampling and an HRTF selection approach. Experimental results show that the proposed method outperforms all three baselines in terms of (LSD) and localisation performance using perceptual models when the input HRTF is sparse (less than 20 measured positions).
-
Conference paperHogg A, Liu H, Mads J, et al., 2023,
Exploring the Impact of Transfer Learning on GAN-Based HRTF Upsampling
, EAA Forum Acusticum, European Congress on Acoustics -
Conference paperSanguedolce G, Naylor PA, Geranmayeh F, 2023, , 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 paperMcKnight S, Hogg AOT, Neo VW, et al., 2022, , APSIPA 2022, Publisher: IEEE, Pages: 394-401
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 paperNeo VW, D'Olne E, Moore AH, et al., 2022, , International Workshop on Acoustic Signal Enhancement (IWAENC), Publisher: IEEE, Pages: 1-5
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 paperNeo VW, Weiss S, McKnight S, et al., 2022, , 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 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: 705-712This 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.
-
Conference paperHogg AOT, Neo VW, Weiss S, et al., 2021, , 2021 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), Publisher: IEEE, Pages: 326-330
-
Conference paperHogg AOT, Evers C, Naylor PA, 2021, , ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Publisher: IEEE, Pages: 26-30
-
Journal articleHogg A, Evers C, Moore A, et al., 2021, , IEEE/ACM Transactions on Audio, Speech and Language Processing, Vol: 29, Pages: 1479-1490, ISSN: 2329-9290
This paper demonstrates how the harmonic structure of voiced speech can be exploited to segment multiple overlapping speakers in a speaker diarization task. We explore how a change in the speaker can be inferred from a change in pitch. We show that voiced harmonics can be useful in detecting when more than one speaker is talking, such as during overlapping speaker activity. A novel system is proposed to track multiple harmonics simultaneously, allowing for the determination of onsets and end-points of a speaker’s utterance in the presence of an additional active speaker. This system is bench-marked against a segmentation system from the literature that employs a bidirectional long short term memory network (BLSTM) approach and requires training. Experimental results highlight that the proposed approach outperforms the BLSTM baseline approach by 12.9% in terms of HIT rate for speaker segmentation. We also show that the estimated pitch tracks of our system can be used as features to the BLSTM to achieve further improvements of 1.21% in terms of coverage and 2.45% in terms of purity
-
Conference paperMcKnight SW, Hogg AOT, Naylor PA, 2021, , 2020 28th European Signal Processing Conference (EUSIPCO), Publisher: IEEE, Pages: 381-385
-
Journal articleSteadman M, Kim C, Lestang J-H, et al., 2019, , Scientific Reports, Vol: 9, ISSN: 2045-2322
Head-related transfer functions (HRTFs) capture the direction-dependant way that sound interacts with the head and torso. In virtual audio systems, which aim to emulate these effects, non-individualized, generic HRTFs are typically used leading to an inaccurate perception of virtual sound location. Training has the potential to exploit the brain’s ability to adapt to these unfamiliar cues. In this study, three virtual sound localization training paradigms were evaluated; one provided simple visual positional confirmation of sound source location, a second introduced game design elements (“gamification”) and a final version additionally utilized head-tracking to provide listeners with experience of relative sound source motion (“active listening”). The results demonstrate a significant effect of training after a small number of short (12-minute) training sessions, which is retained across multiple days. Gamification alone had no significant effect on the efficacy of the training, but active listening resulted in a significantly greater improvements in localization accuracy. In general, improvements in virtual sound localization following training generalized to a second set of non-individualized HRTFs, although some HRTF-specific changes were observed in polar angle judgement for the active listening group. The implications of this on the putative mechanisms of the adaptation process are discussed.
-
Conference paperHogg AOT, Evers C, Naylor PA, 2019, , 2019 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), Publisher: IEEE, Pages: 195-199
-
Conference paperSharma D, Hogg AOT, Wang Y, et al., 2019, , 2019 27th European Signal Processing Conference (EUSIPCO), Publisher: IEEE, Pages: 1-5
-
Conference paperHogg AOT, Evers C, Naylor PA, 2019, , ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Publisher: IEEE, Pages: 5826-5830
-
Journal articleStitt P, Picinali L, Katz BFG, 2019, , Scientific Reports, Vol: 9, Pages: 1-14, ISSN: 2045-2322
This study examines the effect of adaptation to non-ideal auditory localization cues represented by the Head-Related Transfer Function (HRTF) and the retention of training for up to three months after the last session. Continuing from a previous study on rapid non-individual HRTF learning, subjects using non-individual HRTFs were tested alongside control subjects using their own measured HRTFs. Perceptually worst-rated non-individual HRTFs were chosen to represent the worst-case scenario in practice and to allow for maximum potential for improvement. The methodology consisted of a training game and a localization test to evaluate performance carried out over 10 sessions. Sessions 1–4 occurred at 1 week intervals, performed by all subjects. During initial sessions, subjects showed improvement in localization performance for polar error. Following this, half of the subjects stopped the training game element, continuing with only the localization task. The group that continued to train showed improvement, with 3 of 8 subjects achieving group mean polar errors comparable to the control group. The majority of the group that stopped the training game retained their performance attained at the end of session 4. In general, adaptation was found to be quite subject dependent, highlighting the limits of HRTF adaptation in the case of poor HRTF matches. No identifier to predict learning ability was observed.
-
Conference paperMoore A, de Haan JM, Pedersen MS, et al., 2019,
Personalized {HRTF}s for hearing aids
, ELOBES2019 -
Conference paperCuevas-Rodriguez M, Gonzalez-Toledo D, La Rubia-Cuestas ED, et al., 2018,
The 3DTI Toolkit is a standard C++ library for audio spatialisation and simulation using loudspeakers or headphones developed within the 3D Tune-In (3DTI) project (http://www.3d-tune-in.eu), which aims at using 3D sound and simulating hearing loss and hearing aids within virtual environments and games. The Toolkit allows the design and rendering of highly realistic and immersive 3D audio, and the simulation of virtual hearing aid devices and of different typologies of hearing loss. The library includes a real-time 3D binaural audio renderer offering full 3D spatialization based on efficient Head Related Transfer Function (HRTF) convolution, including smooth interpolation among impulse responses, customization of listener head radius and specific simulation of far-distance and near-field effects. In addition, spatial reverberation is simulated in real time using a uniformly partitioned convolution with Binaural Room Impulse Responses (BRIRs) employing a virtual Ambisonic approach. The 3D Tune-In Toolkit includes also a loudspeaker-based spatialiser implemented using Ambisonic encoding/decoding. This poster presents a brief overview of the main features of the Toolkit, which is released open-source under GPL v3 license (the code is available in GitHub https://github.com/3DTune-In/3dti-AudioToolkit).
-
Journal articleBraiman C, Fridman A, Conte MM, et al., 2018, , Current Biology, Vol: 28, Pages: 3833-3839.E3, ISSN: 0960-9822
Recent studies identify severely brain-injured patients with limited or no behavioral responses who successfully perform functional magnetic resonance imaging (fMRI) or electroencephalogram (EEG) mental imagery tasks [1, 2, 3, 4, 5]. Such tasks are cognitively demanding [1]; accordingly, recent studies support that fMRI command following in brain-injured patients associates with preserved cerebral metabolism and preserved sleep-wake EEG [5, 6]. We investigated the use of an EEG response that tracks the natural speech envelope (NSE) of spoken language [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22] in healthy controls and brain-injured patients (vegetative state to emergence from minimally conscious state). As audition is typically preserved after brain injury, auditory paradigms may be preferred in searching for covert cognitive function [23, 24, 25]. NSE measures are obtained by cross-correlating EEG with the NSE. We compared NSE latencies and amplitudes with and without consideration of fMRI assessments. NSE latencies showed significant and progressive delay across diagnostic categories. Patients who could carry out fMRI-based mental imagery tasks showed no statistically significant difference in NSE latencies relative to healthy controls; this subgroup included patients without behavioral command following. The NSE may stratify patients with severe brain injuries and identify those patients demonstrating “cognitive motor dissociation” (CMD) [26] who show only covert evidence of command following utilizing neuroimaging or electrophysiological methods that demand high levels of cognitive function. Thus, the NSE is a passive measure that may provide a useful screening tool to improve detection of covert cognition with fMRI or other methods and improve stratification of patients with disorders of consciousness in research studies.
-
Journal articleSethi S, Ewers R, Jones N, et al., 2018, , Methods in Ecology and Evolution, Vol: 9, Pages: 2383-2387, ISSN: 2041-210X
1. Automated methods of monitoring ecosystems provide a cost-effective way to track changes in natural system's dynamics across temporal and spatial scales. However, methods of recording and storing data captured from the field still require significant manual effort. 2. Here we introduce an open source, inexpensive, fully autonomous ecosystem monitoring unit for capturing and remotely transmitting continuous data streams from field sites over long time-periods. We provide a modular software framework for deploying various sensors, together with implementations to demonstrate proof of concept for continuous audio monitoring and time-lapse photography. 3. We show how our system can outperform comparable technologies for fractions of the cost, provided a local mobile network link is available. The system is robust to unreliable network signals and has been shown to function in extreme environmental conditions, such as in the tropical rainforests of Sabah, Borneo. 4. We provide full details on how to assemble the hardware, and the open-source software. Paired with appropriate automated analysis techniques, this system could provide spatially dense, near real-time, continuous insights into ecosystem and biodiversity dynamics at a low cost.
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.
Contact
For more information about the group, please contact:
Dr Dan Goodman
+44 (0)20 7594 6264
d.goodman@imperial.ac.uk