Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

"speech": models, code, and papers

Utterance partitioning for speaker recognition: an experimental review and analysis with new findings under GMM-SVM framework

May 25, 2021
Nirmalya Sen, Md Sahidullah, Hemant Patil, Shyamal Kumar das Mandal, Sreenivasa Krothapalli Rao, Tapan Kumar Basu

The performance of speaker recognition system is highly dependent on the amount of speech used in enrollment and test. This work presents a detailed experimental review and analysis of the GMM-SVM based speaker recognition system in presence of duration variability. This article also reports a comparison of the performance of GMM-SVM classifier with its precursor technique Gaussian mixture model-universal background model (GMM-UBM) classifier in presence of duration variability. The goal of this research work is not to propose a new algorithm for improving speaker recognition performance in presence of duration variability. However, the main focus of this work is on utterance partitioning (UP), a commonly used strategy to compensate the duration variability issue. We have analysed in detailed the impact of training utterance partitioning in speaker recognition performance under GMM-SVM framework. We further investigate the reason why the utterance partitioning is important for boosting speaker recognition performance. We have also shown in which case the utterance partitioning could be useful and where not. Our study has revealed that utterance partitioning does not reduce the data imbalance problem of the GMM-SVM classifier as claimed in earlier study. Apart from these, we also discuss issues related to the impact of parameters such as number of Gaussians, supervector length, amount of splitting required for obtaining better performance in short and long duration test conditions from speech duration perspective. We have performed the experiments with telephone speech from POLYCOST corpus consisting of 130 speakers.

* International Journal of Speech Technology, Springer Verlag, In press 

  Access Paper or Ask Questions

Multilingual training set selection for ASR in under-resourced Malian languages

Aug 13, 2021
Ewald van der Westhuizen, Trideba Padhi, Thomas Niesler

We present first speech recognition systems for the two severely under-resourced Malian languages Bambara and Maasina Fulfulde. These systems will be used by the United Nations as part of a monitoring system to inform and support humanitarian programmes in rural Africa. We have compiled datasets in Bambara and Maasina Fulfulde, but since these are very small, we take advantage of six similarly under-resourced datasets in other languages for multilingual training. We focus specifically on the best composition of the multilingual pool of speech data for multilingual training. We find that, although maximising the training pool by including all six additional languages provides improved speech recognition in both target languages, substantially better performance can be achieved by a more judicious choice. Our experiments show that the addition of just one language provides best performance. For Bambara, this additional language is Maasina Fulfulde, and its introduction leads to a relative word error rate reduction of 6.7%, as opposed to a 2.4% relative reduction achieved when pooling all six additional languages. For the case of Maasina Fulfulde, best performance was achieved when adding only Luganda, leading to a relative word error rate improvement of 9.4% as opposed to a 3.9% relative improvement when pooling all six languages. We conclude that careful selection of the out-of-language data is worthwhile for multilingual training even in highly under-resourced settings, and that the general assumption that more data is better does not always hold.

* 12 pages, 4 figures, Accepted for presentation at SPECOM 2021 

  Access Paper or Ask Questions

Reverberant Sound Localization with a Robot Head Based on Direct-Path Relative Transfer Function

Dec 07, 2020
Xiaofei Li, Laurent Girin, Fabien Badeig, Radu Horaud

This paper addresses the problem of sound-source localization (SSL) with a robot head, which remains a challenge in real-world environments. In particular we are interested in locating speech sources, as they are of high interest for human-robot interaction. The microphone-pair response corresponding to the direct-path sound propagation is a function of the source direction. In practice, this response is contaminated by noise and reverberations. The direct-path relative transfer function (DP-RTF) is defined as the ratio between the direct-path acoustic transfer function (ATF) of the two microphones, and it is an important feature for SSL. We propose a method to estimate the DP-RTF from noisy and reverberant signals in the short-time Fourier transform (STFT) domain. First, the convolutive transfer function (CTF) approximation is adopted to accurately represent the impulse response of the microphone array, and the first coefficient of the CTF is mainly composed of the direct-path ATF. At each frequency, the frame-wise speech auto- and cross-power spectral density (PSD) are obtained by spectral subtraction. Then a set of linear equations is constructed by the speech auto- and cross-PSD of multiple frames, in which the DP-RTF is an unknown variable, and is estimated by solving the equations. Finally, the estimated DP-RTFs are concatenated across frequencies and used as a feature vector for SSL. Experiments with a robot, placed in various reverberant environments, show that the proposed method outperforms two state-of-the-art methods.

* IEEE/RSJ International Conference on Intelligent Robots and Systems, 

  Access Paper or Ask Questions

MESH2IR: Neural Acoustic Impulse Response Generator for Complex 3D Scenes

May 18, 2022
Anton Ratnarajah, Zhenyu Tang, Rohith Chandrashekar Aralikatti, Dinesh Manocha

We propose a mesh-based neural network (MESH2IR) to generate acoustic impulse responses (IRs) for indoor 3D scenes represented using a mesh. The IRs are used to create a high-quality sound experience in interactive applications and audio processing. Our method can handle input triangular meshes with arbitrary topologies (2K - 3M triangles). We present a novel training technique to train MESH2IR using energy decay relief and highlight its benefits. We also show that training MESH2IR on IRs preprocessed using our proposed technique significantly improves the accuracy of IR generation. We reduce the non-linearity in the mesh space by transforming 3D scene meshes to latent space using a graph convolution network. Our MESH2IR is more than 200 times faster than a geometric acoustic algorithm on a CPU and can generate more than 10,000 IRs per second on an NVIDIA GeForce RTX 2080 Ti GPU for a given furnished indoor 3D scene. The acoustic metrics are used to characterize the acoustic environment. We show that the acoustic metrics of the IRs predicted from our MESH2IR match the ground truth with less than 10% error. We also highlight the benefits of MESH2IR on audio and speech processing applications such as speech dereverberation and speech separation. To the best of our knowledge, ours is the first neural-network-based approach to predict IRs from a given 3D scene mesh in real-time.

* More results and source code is available at 

  Access Paper or Ask Questions

Operationalizing the legal concept of 'Incitement to Hatred' as an NLP task

Apr 07, 2020
Frederike Zufall, Huangpan Zhang, Katharina Kloppenborg, Torsten Zesch

Hate speech detection or offensive language detection are well-established but controversial NLP tasks. There is no denying the temptation to use them for law enforcement or by private actors to censor, delete, or punish online statements. However, given the importance of freedom of expression for the public discourse in a democracy, determining statements that would potentially be subject to these measures requires a legal justification that outweighs the right to free speech in the respective case. The legal concept of 'incitement to hatred' answers this question by preventing discrimination against and segregation of a target group, thereby ensuring the members' acceptance as equal in a society - likewise a prerequisite for democracy. In this paper, we pursue these questions based on the criminal offense of 'incitement to hatred' in {\S} 130 of the German Criminal Code along with the underlying EU Council Framework Decision. Under the German Network Enforcement Act, social media providers are subject to a direct obligation to delete postings violating this offense. We take this as a use case to study the transition from the ill-defined concepts of hate speech or offensive language which are usually used in NLP to an operationalization of an actual legally binding obligation. We first translate the legal assessment into a series of binary decisions and then collect, annotate, and analyze a dataset according to our annotation scheme. Finally, we translate each of the legal decisions into an NLP task based on the annotated data. In this way, we ultimately also explore the extent to which the underlying value-based decisions could be carried over to NLP.

  Access Paper or Ask Questions

Initial investigation of an encoder-decoder end-to-end TTS framework using marginalization of monotonic hard latent alignments

Aug 30, 2019
Yusuke Yasuda, Xin Wang, Junichi Yamagishi

End-to-end text-to-speech (TTS) synthesis is a method that directly converts input text to output acoustic features using a single network. A recent advance of end-to-end TTS is due to a key technique called attention mechanisms, and all successful methods proposed so far have been based on soft attention mechanisms. However, although network structures are becoming increasingly complex, end-to-end TTS systems with soft attention mechanisms may still fail to learn and to predict accurate alignment between the input and output. This may be because the soft attention mechanisms are too flexible. Therefore, we propose an approach that has more explicit but natural constraints suitable for speech signals to make alignment learning and prediction of end-to-end TTS systems more robust. The proposed system, with the constrained alignment scheme borrowed from segment-to-segment neural transduction (SSNT), directly calculates the joint probability of acoustic features and alignment given an input text. The alignment is designed to be hard and monotonically increase by considering the speech nature, and it is treated as a latent variable and marginalized during training. During prediction, both the alignment and acoustic features can be generated from the probabilistic distributions. The advantages of our approach are that we can simplify many modules for the soft attention and that we can train the end-to-end TTS model using a single likelihood function. As far as we know, our approach is the first end-to-end TTS without a soft attention mechanism.

* To be appeared at SSW10 

  Access Paper or Ask Questions

Approximations to the MMI criterion and their effect on lattice-based MMI

Feb 03, 2010
Steven Wegmann

Maximum mutual information (MMI) is a model selection criterion used for hidden Markov model (HMM) parameter estimation that was developed more than twenty years ago as a discriminative alternative to the maximum likelihood criterion for HMM-based speech recognition. It has been shown in the speech recognition literature that parameter estimation using the current MMI paradigm, lattice-based MMI, consistently outperforms maximum likelihood estimation, but this is at the expense of undesirable convergence properties. In particular, recognition performance is sensitive to the number of times that the iterative MMI estimation algorithm, extended Baum-Welch, is performed. In fact, too many iterations of extended Baum-Welch will lead to degraded performance, despite the fact that the MMI criterion improves at each iteration. This phenomenon is at variance with the analogous behavior of maximum likelihood estimation -- at least for the HMMs used in speech recognition -- and it has previously been attributed to `over fitting'. In this paper, we present an analysis of lattice-based MMI that demonstrates, first of all, that the asymptotic behavior of lattice-based MMI is much worse than was previously understood, i.e. it does not appear to converge at all, and, second of all, that this is not due to `over fitting'. Instead, we demonstrate that the `over fitting' phenomenon is the result of standard methodology that exacerbates the poor behavior of two key approximations in the lattice-based MMI machinery. We also demonstrate that if we modify the standard methodology to improve the validity of these approximations, then the convergence properties of lattice-based MMI become benign without sacrificing improvements to recognition accuracy.

  Access Paper or Ask Questions

Automatic audiovisual synchronisation for ultrasound tongue imaging

May 31, 2021
Aciel Eshky, Joanne Cleland, Manuel Sam Ribeiro, Eleanor Sugden, Korin Richmond, Steve Renals

Ultrasound tongue imaging is used to visualise the intra-oral articulators during speech production. It is utilised in a range of applications, including speech and language therapy and phonetics research. Ultrasound and speech audio are recorded simultaneously, and in order to correctly use this data, the two modalities should be correctly synchronised. Synchronisation is achieved using specialised hardware at recording time, but this approach can fail in practice resulting in data of limited usability. In this paper, we address the problem of automatically synchronising ultrasound and audio after data collection. We first investigate the tolerance of expert ultrasound users to synchronisation errors in order to find the thresholds for error detection. We use these thresholds to define accuracy scoring boundaries for evaluating our system. We then describe our approach for automatic synchronisation, which is driven by a self-supervised neural network, exploiting the correlation between the two signals to synchronise them. We train our model on data from multiple domains with different speaker characteristics, different equipment, and different recording environments, and achieve an accuracy >92.4% on held-out in-domain data. Finally, we introduce a novel resource, the Cleft dataset, which we gathered with a new clinical subgroup and for which hardware synchronisation proved unreliable. We apply our model to this out-of-domain data, and evaluate its performance subjectively with expert users. Results show that users prefer our model's output over the original hardware output 79.3% of the time. Our results demonstrate the strength of our approach and its ability to generalise to data from new domains.

* 18 pages, 10 figures. Manuscript accepted at Speech Communication 

  Access Paper or Ask Questions

Auditory System for a Mobile Robot

Feb 22, 2016
Jean-Marc Valin

In this thesis, we propose an artificial auditory system that gives a robot the ability to locate and track sounds, as well as to separate simultaneous sound sources and recognising simultaneous speech. We demonstrate that it is possible to implement these capabilities using an array of microphones, without trying to imitate the human auditory system. The sound source localisation and tracking algorithm uses a steered beamformer to locate sources, which are then tracked using a multi-source particle filter. Separation of simultaneous sound sources is achieved using a variant of the Geometric Source Separation (GSS) algorithm, combined with a multi-source post-filter that further reduces noise, interference and reverberation. Speech recognition is performed on separated sources, either directly or by using Missing Feature Theory (MFT) to estimate the reliability of the speech features. The results obtained show that it is possible to track up to four simultaneous sound sources, even in noisy and reverberant environments. Real-time control of the robot following a sound source is also demonstrated. The sound source separation approach we propose is able to achieve a 13.7 dB improvement in signal-to-noise ratio compared to a single microphone when three speakers are present. In these conditions, the system demonstrates more than 80% accuracy on digit recognition, higher than most human listeners could obtain in our small case study when recognising only one of these sources. All these new capabilities will allow humans to interact more naturally with a mobile robot in real life settings.

* 120 pages, PhD thesis, University of Sherbrooke, 2005 

  Access Paper or Ask Questions

Unsupervised Discovery of Linguistic Structure Including Two-level Acoustic Patterns Using Three Cascaded Stages of Iterative Optimization

Sep 07, 2015
Cheng-Tao Chung, Chun-an Chan, Lin-shan Lee

Techniques for unsupervised discovery of acoustic patterns are getting increasingly attractive, because huge quantities of speech data are becoming available but manual annotations remain hard to acquire. In this paper, we propose an approach for unsupervised discovery of linguistic structure for the target spoken language given raw speech data. This linguistic structure includes two-level (subword-like and word-like) acoustic patterns, the lexicon of word-like patterns in terms of subword-like patterns and the N-gram language model based on word-like patterns. All patterns, models, and parameters can be automatically learned from the unlabelled speech corpus. This is achieved by an initialization step followed by three cascaded stages for acoustic, linguistic, and lexical iterative optimization. The lexicon of word-like patterns defines allowed consecutive sequence of HMMs for subword-like patterns. In each iteration, model training and decoding produces updated labels from which the lexicon and HMMs can be further updated. In this way, model parameters and decoded labels are respectively optimized in each iteration, and the knowledge about the linguistic structure is learned gradually layer after layer. The proposed approach was tested in preliminary experiments on a corpus of Mandarin broadcast news, including a task of spoken term detection with performance compared to a parallel test using models trained in a supervised way. Results show that the proposed system not only yields reasonable performance on its own, but is also complimentary to existing large vocabulary ASR systems.

* Accepted by ICASSP 2013 

  Access Paper or Ask Questions