Time-frequency spectral estimation of speech

In recent years there has been a growing interest amongst the speech research community into the use of spectral estimators which circumvent the traditional quasi-stationary assumption and provide greater time-frequency (t-f) resolution than conventional spectral estimators, such as the short time Fourier power spectrum (STFPS). One distribution in particular, the Wigner distribution (WD), has attracted considerable interest. However, experimental studies have indicated that, despite its improved t-f resolution, employing the WD as the front end of speech recognition system actually reduces recognition performance; only by explicitly re-introducing t-f smoothing into the WD are recognition rates improved. In this paper we provide an explanation for these findings. By treating the spectral estimation problem as one of optimization of a bias variance trade off, we show why additional t-f smoothing improves recognition rates, despite reducing the t-f resolution of the spectral estimator. A practical adaptive smoothing algorithm is presented, whicy attempts to match the degree of smoothing introduced into the WD with the time varying quasi-stationary regions within the speech waveform. The recognition performance of the resulting adaptively smoothed estimator is found to be comparable to that of conventional filterbank estimators, yet the average temporal sampling rate of the resulting spectral vectors is reduced by around a factor of 10. © 1992.

Unsupervised intra-lingual and cross-lingual speaker adaptation for HMM-based speech synthesis using two-pass decision tree construction

Hidden Markov model (HMM)-based speech synthesis systems possess several advantages over concatenative synthesis systems. One such advantage is the relative ease with which HMM-based systems are adapted to speakers not present in the training dataset. Speaker adaptation methods used in the field of HMM-based automatic speech recognition (ASR) are adopted for this task. In the case of unsupervised speaker adaptation, previous work has used a supplementary set of acoustic models to estimate the transcription of the adaptation data. This paper first presents an approach to the unsupervised speaker adaptation task for HMM-based speech synthesis models which avoids the need for such supplementary acoustic models. This is achieved by defining a mapping between HMM-based synthesis models and ASR-style models, via a two-pass decision tree construction process. Second, it is shown that this mapping also enables unsupervised adaptation of HMM-based speech synthesis models without the need to perform linguistic analysis of the estimated transcription of the adaptation data. Third, this paper demonstrates how this technique lends itself to the task of unsupervised cross-lingual adaptation of HMM-based speech synthesis models, and explains the advantages of such an approach. Finally, listener evaluations reveal that the proposed unsupervised adaptation methods deliver performance approaching that of supervised adaptation.

Word boundary modelling and full covariance gaussians for Arabic Speech-to-Text systems

This paper describes recent improvements to the Cambridge Arabic Large Vocabulary Continuous Speech Recognition (LVCSR) Speech-to-Text (STT) system. It is shown that wordboundary context markers provide a powerful method to enhance graphemic systems by implicit phonetic information, improving the modelling capability of graphemic systems. In addition, a robust technique for full covariance Gaussian modelling in the Minimum Phone Error (MPE) training framework is introduced. This reduces the full covariance training to a diagonal covariance training problem, thereby solving related robustness problems. The full system results show that the combined use of these and other techniques within a multi-branch combination framework reduces the Word Error Rate (WER) of the complete system by up to 5.9% relative. Copyright © 2011 ISCA.

Vowel normalisation: Time-domain processing of the internal dynamics of speech

Human listeners can identify vowels regardless of speaker size, although the sound waves for an adult and a child speaking the ’same’ vowel would differ enormously. The differences are mainly due to the differences in vocal tract length (VTL) and glottal pulse rate (GPR) which are both related to body size. Automatic speech recognition machines are notoriously bad at understanding children if they have been trained on the speech of an adult. In this paper, we propose that the auditory system adapts its analysis of speech sounds, dynamically and automatically to the GPR and VTL of the speaker on a syllable-to-syllable basis. We illustrate how this rapid adaptation might be performed with the aid of a computational version of the auditory image model, and we propose that an auditory preprocessor of this form would improve the robustness of speech recognisers.

Elicitation techniques for the evaluation of speech under stress: Towards a framework for use in a controlled environment

Existing work in Computer Science and Electronic Engineering demonstrates that Digital Signal Processing techniques can effectively identify the presence of stress in the speech signal. These techniques use datasets containing real or actual stress samples i.e. real-life stress such as 911 calls and so on. Studies that use simulated or laboratory-induced stress have been less successful and inconsistent. Pervasive, ubiquitous computing is increasingly moving towards voice-activated and voice-controlled systems and devices. Speech recognition and speaker identification algorithms will have to improve and take emotional speech into account. Modelling the influence of stress on speech and voice is of interest to researchers from many different disciplines including security, telecommunications, psychology, speech science, forensics and Human Computer Interaction (HCI). The aim of this work is to assess the impact of moderate stress on the speech signal. In order to do this, a dataset of laboratory-induced stress is required. While attempting to build this dataset it became apparent that reliably inducing measurable stress in a controlled environment, when speech is a requirement, is a challenging task. This work focuses on the use of a variety of stressors to elicit a stress response during tasks that involve speech content. Biosignal analysis (commercial Brain Computer Interfaces, eye tracking and skin resistance) is used to verify and quantify the stress response, if any. This thesis explains the basis of the author’s hypotheses on the elicitation of affectively-toned speech and presents the results of several studies carried out throughout the PhD research period. These results show that the elicitation of stress, particularly the induction of affectively-toned speech, is not a simple matter and that many modulating factors influence the stress response process. A model is proposed to reflect the author’s hypothesis on the emotional response pathways relating to the elicitation of stress with a required speech content. Finally the author provides guidelines and recommendations for future research on speech under stress. Further research paths are identified and a roadmap for future research in this area is defined.

On-line Emotion Recognition in a 3-D Activation-Valence-Time Continuum using Acoustic and Linguistic Cues

For many applications of emotion recognition, such as virtual agents, the system must select responses while the user is speaking. This requires reliable on-line recognition of the user’s affect. However most emotion recognition systems are based on turnwise processing. We present a novel approach to on-line emotion recognition from speech using Long Short-Term Memory Recurrent Neural Networks. Emotion is recognised frame-wise in a two-dimensional valence-activation continuum. In contrast to current state-of-the-art approaches, recognition is performed on low-level signal frames, similar to those used for speech recognition. No statistical functionals are applied to low-level feature contours. Framing at a higher level is therefore unnecessary and regression outputs can be produced in real-time for every low-level input frame. We also investigate the benefits of including linguistic features on the signal frame level obtained by a keyword spotter.

Creating a spontaneous conversational speech corpus

Speech recognition and language analysis of spontaneous speech arising in naturally spoken conversations are becoming the subject of much research. However, there is a shortage of spontaneous speech corpora that are freely available for academics. We therefore undertook the building of a natural conversation speech database, recording over 200 hours of conversations in English by over 600 local university students. With few exceptions, the students used their own cell phones from their own rooms or homes to speak to one another, and they were permitted to speak on any topic they chose. Although they knew that they were being recorded and that they would receive a small payment, their conversations in the corpus are probably very close to being natural and spontaneous. This paper describes a detailed case study of the problems we faced and the methods we used to make the recordings and control the collection of these social science data on a limited budget.

CLOSE—A Data-Driven Approach to Speech Separation

This paper studies single-channel speech separation, assuming unknown, arbitrary temporal dynamics for the speech signals to be separated. A data-driven approach is described, which matches each mixed speech segment against a composite training segment to separate the underlying clean speech segments. To advance the separation accuracy, the new approach seeks and separates the longest mixed speech segments with matching composite training segments. Lengthening the mixed speech segments to match reduces the uncertainty of the constituent training segments, and hence the error of separation. For convenience, we call the new approach Composition of Longest Segments, or CLOSE. The CLOSE method includes a data-driven approach to model long-range temporal dynamics of speech signals, and a statistical approach to identify the longest mixed speech segments with matching composite training segments. Experiments are conducted on the Wall Street Journal database, for separating mixtures of two simultaneous large-vocabulary speech utterances spoken by two different speakers. The results are evaluated using various objective and subjective measures, including the challenge of large-vocabulary continuous speech recognition. It is shown that the new separation approach leads to significant improvement in all these measures.

Combining missing-feature theory, speech enhancement, and speaker-dependent/-independent modeling for speech separation

This paper considers the separation and recognition of overlapped speech sentences assuming single-channel observation. A system based on a combination of several different techniques is proposed. The system uses a missing-feature approach for improving crosstalk/noise robustness, a Wiener filter for speech enhancement, hidden Markov models for speech reconstruction, and speaker-dependent/-independent modeling for speaker and speech recognition. We develop the system on the Speech Separation Challenge database, involving a task of separating and recognizing two mixing sentences without assuming advanced knowledge about the identity of the speakers nor about the signal-to-noise ratio. The paper is an extended version of a previous conference paper submitted for the challenge.

An iterative longest matching segment approach to speech enhancement with additive noise and channel distortion

This paper presents a new approach to speech enhancement from single-channel measurements involving both noise and channel distortion (i.e., convolutional noise), and demonstrates its applications for robust speech recognition and for improving noisy speech quality. The approach is based on finding longest matching segments (LMS) from a corpus of clean, wideband speech. The approach adds three novel developments to our previous LMS research. First, we address the problem of channel distortion as well as additive noise. Second, we present an improved method for modeling noise for speech estimation. Third, we present an iterative algorithm which updates the noise and channel estimates of the corpus data model. In experiments using speech recognition as a test with the Aurora 4 database, the use of our enhancement approach as a preprocessor for feature extraction significantly improved the performance of a baseline recognition system. In another comparison against conventional enhancement algorithms, both the PESQ and the segmental SNR ratings of the LMS algorithm were superior to the other methods for noisy speech enhancement.

Speech Enhancement from Additive Noise and Channel Distortion - a Corpus-Based Approach

This paper presents a new approach to single-channel speech enhancement involving both noise and channel distortion (i.e., convolutional noise). The approach is based on finding longest matching segments (LMS) from a corpus of clean, wideband speech. The approach adds three novel developments to our previous LMS research. First, we address the problem of channel distortion as well as additive noise. Second, we present an improved method for modeling noise. Third, we present an iterative algorithm for improved speech estimates. In experiments using speech recognition as a test with the Aurora 4 database, the use of our enhancement approach as a preprocessor for feature extraction significantly improved the performance of a baseline recognition system. In another comparison against conventional enhancement algorithms, both the PESQ and the segmental SNR ratings of the LMS algorithm were superior to the other methods for noisy speech enhancement. Index Terms: corpus-based speech model, longest matching segment, speech enhancement, speech recognition

Adaptive modeling and high quality spectral estimation for speech enhancement

In this work an adaptive modeling and spectral estimation scheme based on a dual Discrete Kalman Filtering (DKF) is proposed for speech enhancement. Both speech and noise signals are modeled by an autoregressive structure which provides an underlying time frame dependency and improves time-frequency resolution. The model parameters are arranged to obtain a combined state-space model and are also used to calculate instantaneous power spectral density estimates. The speech enhancement is performed by a dual discrete Kalman filter that simultaneously gives estimates for the models and the signals. This approach is particularly useful as a pre-processing module for parametric based speech recognition systems that rely on spectral time dependent models. The system performance has been evaluated by a set of human listeners and by spectral distances. In both cases the use of this pre-processing module has led to improved results.

A Hybrid Architecture for Recognising Speech Signals in Malayalam

Speech is the primary, most prominent and convenient means of communication in audible language. Through speech, people can express their thoughts, feelings or perceptions by the articulation of words. Human speech is a complex signal which is non stationary in nature. It consists of immensely rich information about the words spoken, accent, attitude of the speaker, expression, intention, sex, emotion as well as style. The main objective of Automatic Speech Recognition (ASR) is to identify whatever people speak by means of computer algorithms. This enables people to communicate with a computer in a natural spoken language. Automatic recognition of speech by machines has been one of the most exciting, significant and challenging areas of research in the field of signal processing over the past five to six decades. Despite the developments and intensive research done in this area, the performance of ASR is still lower than that of speech recognition by humans and is yet to achieve a completely reliable performance level. The main objective of this thesis is to develop an efficient speech recognition system for recognising speaker independent isolated words in Malayalam.

Segmentation and Alignment of Speech and Sketching in a Design Environment

Sketches are commonly used in the early stages of design. Our previous system allows users to sketch mechanical systems that the computer interprets. However, some parts of the mechanical system might be too hard or too complicated to express in the sketch. Adding speech recognition to create a multimodal system would move us toward our goal of creating a more natural user interface. This thesis examines the relationship between the verbal and sketch input, particularly how to segment and align the two inputs. Toward this end, subjects were recorded while they sketched and talked. These recordings were transcribed, and a set of rules to perform segmentation and alignment was created. These rules represent the knowledge that the computer needs to perform segmentation and alignment. The rules successfully interpreted the 24 data sets that they were given.

Perceptual Evaluation of Video-Realistic Speech

abstract With many visual speech animation techniques now available, there is a clear need for systematic perceptual evaluation schemes. We describe here our scheme and its application to a new video-realistic (potentially indistinguishable from real recorded video) visual-speech animation system, called Mary 101. Two types of experiments were performed: a) distinguishing visually between real and synthetic image- sequences of the same utterances, ("Turing tests") and b) gauging visual speech recognition by comparing lip-reading performance of the real and synthetic image-sequences of the same utterances ("Intelligibility tests"). Subjects that were presented randomly with either real or synthetic image-sequences could not tell the synthetic from the real sequences above chance level. The same subjects when asked to lip-read the utterances from the same image-sequences recognized speech from real image-sequences significantly better than from synthetic ones. However, performance for both, real and synthetic, were at levels suggested in the literature on lip-reading. We conclude from the two experiments that the animation of Mary 101 is adequate for providing a percept of a talking head. However, additional effort is required to improve the animation for lip-reading purposes like rehabilitation and language learning. In addition, these two tasks could be considered as explicit and implicit perceptual discrimination tasks. In the explicit task (a), each stimulus is classified directly as a synthetic or real image-sequence by detecting a possible difference between the synthetic and the real image-sequences. The implicit perceptual discrimination task (b) consists of a comparison between visual recognition of speech of real and synthetic image-sequences. Our results suggest that implicit perceptual discrimination is a more sensitive method for discrimination between synthetic and real image-sequences than explicit perceptual discrimination.