The use of phase in complex spectrum subtraction for robust speech recognition

In this paper we propose a new method for utilising phase information by complementing it with traditional magnitude-only spectral subtraction speech enhancement through Complex Spectrum Subtraction (CSS). The proposed approach has the following advantages over traditional magnitude-only spectral subtraction: (a) it introduces complementary information to the enhancement algorithm; (b) it reduces the total number of algorithmic parameters, and; (c) is designed for improving clean speech magnitude spectra and is therefore suitable for both automatic speech recognition (ASR) and speech perception applications. Oracle-based ASR experiments verify this approach, showing an average of 20% relative word accuracy improvements when accurate estimates of the phase spectrum are available. Based on sinusoidal analysis and assuming stationarity between observations (which is shown to be better approximated as the frame rate is increased), this paper also proposes a novel method for acquiring the phase information called Phase Estimation via Delay Projection (PEDEP). Further oracle ASR experiments validate the potential for the proposed PEDEP technique in ideal conditions. Realistic implementation of CSS with PEDEP shows performance comparable to state of the art spectral subtraction techniques in a range of 15-20 dB signal-to-noise ratio environments. These results clearly demonstrate the potential for using phase spectra in spectral subtractive enhancement applications, and at the same time highlight the need for deriving more accurate phase estimates in a wider range of noise conditions.

Improving visual noise insensitivity in small vocabulary audio visual speech recognition applications

Visual noise insensitivity is important to audio visual speech recognition (AVSR). Visual noise can take on a number of forms such as varying frame rate, occlusion, lighting or speaker variabilities. The use of a high dimensional secondary classifier on the word likelihood scores from both the audio and video modalities is investigated for the purposes of adaptive fusion. Preliminary results are presented demonstrating performance above the catastrophic fusion boundary for our confidence measure irrespective of the type of visual noise presented to it. Our experiments were restricted to small vocabulary applications.

fMRI evidence of word frequency and strength effects during episodic memory encoding

Word frequency (WF) and strength effects are two important phenomena associated with episodic memory. The former refers to the superior hit-rate (HR) for low (LF) compared to high frequency (HF) words in recognition memory, while the latter describes the incremental effect(s) upon HRs associated with repeating an item at study. Using the "subsequent memory" method with event-related fMRI, we tested the attention-at-encoding (AE) [M. Glanzer, J.K. Adams, The mirror effect in recognition memory: data and theory, J. Exp. Psychol.: Learn Mem. Cogn. 16 (1990) 5-16] explanation of the WF effect. In addition to investigating encoding strength, we addressed if study involves accessing prior representations of repeated items via the same mechanism as that at test [J.L. McClelland, M. Chappell, Familiarity breeds differentiation: a subjective-likelihood approach to the effects of experience in recognition memory, Psychol. Rev. 105 (1998) 724-760], entailing recollection [K.J. Malmberg, J.E. Holden, R.M. Shiffrin, Modeling the effects of repetitions, similarity, and normative word frequency on judgments of frequency and recognition memory, J. Exp. Psychol.: Learn Mem. Cogn. 30 (2004) 319-331] and whether less processing effort is entailed for encoding each repetition [M. Cary, L.M. Reder, A dual-process account of the list-length and strength-based mirror effects in recognition, J. Mem. Lang. 49 (2003) 231-248]. The increased BOLD responses observed in the left inferior prefrontal cortex (LIPC) for the WF effect provide support for an AE account. Less effort does appear to be required for encoding each repetition of an item, as reduced BOLD responses were observed in the LIPC and left lateral temporal cortex; both regions demonstrated increased responses in the conventional subsequent memory analysis. At test, a left lateral parietal BOLD response was observed for studied versus unstudied items, while only medial parietal activity was observed for repeated items at study, indicating that accessing prior representations at encoding does not necessarily occur via the same mechanism as that at test, and is unlikely to involve a conscious recall-like process such as recollection. This information may prove useful for constraining cognitive theories of episodic memory.

Channel selection in the short-time modulation domain for distant speech recognition

Automatic speech recognition from multiple distant micro- phones poses significant challenges because of noise and reverberations. The quality of speech acquisition may vary between microphones because of movements of speakers and channel distortions. This paper proposes a channel selection approach for selecting reliable channels based on selection criterion operating in the short-term modulation spectrum domain. The proposed approach quantifies the relative strength of speech from each microphone and speech obtained from beamforming modulations. The new technique is compared experimentally in the real reverb conditions in terms of perceptual evaluation of speech quality (PESQ) measures and word error rate (WER). Overall improvement in recognition rate is observed using delay-sum and superdirective beamformers compared to the case when the channel is selected randomly using circular microphone arrays.

One Word, One Body, One Voice : Studies in Apophatic Theology and Christocentric Anthropology in Gregory of Nyssa

The thesis consists of five international congress papers and a summary with an introduction. The overarching aim of the studies and the summary is to examine the inner coherency of the theological and anthropological thinking of Gregory of Nyssa (331-395). To the issue is applied an "apophatic approach" with a "Christological focus". It is suggested that the coherency is to be found from the Christological concept of unity between "true God" and "true man" in the one person of Jesus Christ. Gregory is among the first to make a full recognition of two natures of Christ, and to use this recognition systematically in his writings. The aim of the studies is pursued by the method of "identification", a combination of the modern critical "problematic method" and Gregory's own aphairetic method of "following" (akolouthia). The preoccupation with issues relating to the so-called Hellenization of Christianity in the patristic era was strong in the twentieth-century Gregory scholarship. The most discussed questions have been the Greek influence in his thought and his philosophical sources. In the five articles of the thesis it is examined how Gregory's thinking stands in its own right. The manifestly apophatic character of his theological thinking is made a part of the method of examining his thought according to the principles of his own method of following. The basic issue concerning the relation of theology and anthropology is discussed in the contexts of his central Trinitarian, anhtropological, Christological and eschatological sources. In the summary the Christocentric integration of Gregory's thinking is discussed also in relation to the issue of the alledged Hellenization. The main conclusion of the thesis concerns the concept of theology in Gregory. It is not indebted to the classical concept of theology as metaphysics or human speculation of God. Instead, it is founded to the traditional Judeo-Christian idea of God who speaks with his people face to face. In Gregory, theologia connotes the oikonomia of God's self-revelation. It may be regarded as the state of constant expression of love between the Creator and his created image. In theology, the human person becomes an image of the Word by which the Father expresses his love to "man" whom he loves as his own Son. Eventually the whole humankind, as one, gives the divine Word a physical - audible and sensible - Body. Humankind then becomes what theology is. The whole humanity expresses divine love by manifesting Christ in words and deeds, singing in one voice to the glory of the Father, the Son and the Holy Spirit.

Machine recognition of printed Kannada text

This paper presents the design of a full fledged OCR system for printed Kannada text. The machine recognition of Kannada characters is difficult due to similarity in the shapes of different characters, script complexity and non-uniqueness in the representation of diacritics. The document image is subject to line segmentation, word segmentation and zone detection. From the zonal information, base characters, vowel modifiers and consonant conjucts are separated. Knowledge based approach is employed for recognizing the base characters. Various features are employed for recognising the characters. These include the coefficients of the Discrete Cosine Transform, Discrete Wavelet Transform and Karhunen-Louve Transform. These features are fed to different classifiers. Structural features are used in the subsequent levels to discriminate confused characters. Use of structural features, increases recognition rate from 93% to 98%. Apart from the classical pattern classification technique of nearest neighbour, Artificial Neural Network (ANN) based classifiers like Back Propogation and Radial Basis Function (RBF) Networks have also been studied. The ANN classifiers are trained in supervised mode using the transform features. Highest recognition rate of 99% is obtained with RBF using second level approximation coefficients of Haar wavelets as the features on presegmented base characters.

Online Handwritten Kannada Word Recognizer with Unrestricted Vocabulary

In this paper, we propose a novel heuristic approach to segment recognizable symbols from online Kannada word data and perform recognition of the entire word. Two different estimates of first derivative are extracted from the preprocessed stroke groups and used as features for classification. Estimate 2 proved better resulting in 88% accuracy, which is 3% more than that achieved with estimate 1. Classification is performed by statistical dynamic space warping (SDSW) classifier which uses X, Y co-ordinates and their first derivatives as features. Classifier is trained with data from 40 writers. 295 classes are handled covering Kannada aksharas, with Kannada numerals, Indo-Arabic numerals, punctuations and other special symbols like $ and #. Classification accuracies obtained are 88% at the akshara level and 80% at the word level, which shows the scope for further improvement in segmentation algorithm

Fast Likelihood Computation in Speech Recognition using Matrices

Acoustic modeling using mixtures of multivariate Gaussians is the prevalent approach for many speech processing problems. Computing likelihoods against a large set of Gaussians is required as a part of many speech processing systems and it is the computationally dominant phase for Large Vocabulary Continuous Speech Recognition (LVCSR) systems. We express the likelihood computation as a multiplication of matrices representing augmented feature vectors and Gaussian parameters. The computational gain of this approach over traditional methods is by exploiting the structure of these matrices and efficient implementation of their multiplication. In particular, we explore direct low-rank approximation of the Gaussian parameter matrix and indirect derivation of low-rank factors of the Gaussian parameter matrix by optimum approximation of the likelihood matrix. We show that both the methods lead to similar speedups but the latter leads to far lesser impact on the recognition accuracy. Experiments on 1,138 work vocabulary RM1 task and 6,224 word vocabulary TIMIT task using Sphinx 3.7 system show that, for a typical case the matrix multiplication based approach leads to overall speedup of 46 % on RM1 task and 115 % for TIMIT task. Our low-rank approximation methods provide a way for trading off recognition accuracy for a further increase in computational performance extending overall speedups up to 61 % for RM1 and 119 % for TIMIT for an increase of word error rate (WER) from 3.2 to 3.5 % for RM1 and for no increase in WER for TIMIT. We also express pairwise Euclidean distance computation phase in Dynamic Time Warping (DTW) in terms of matrix multiplication leading to saving of approximately of computational operations. In our experiments using efficient implementation of matrix multiplication, this leads to a speedup of 5.6 in computing the pairwise Euclidean distances and overall speedup up to 3.25 for DTW.

Improved recognition of aged Kannada documents by effective segmentation of merged characters

In optical character recognition of very old books, the recognition accuracy drops mainly due to the merging or breaking of characters. In this paper, we propose the first algorithm to segment merged Kannada characters by using a hypothesis to select the positions to be cut. This method searches for the best possible positions to segment, by taking into account the support vector machine classifier's recognition score and the validity of the aspect ratio (width to height ratio) of the segments between every pair of cut positions. The hypothesis to select the cut position is based on the fact that a concave surface exists above and below the touching portion. These concave surfaces are noted down by tracing the valleys in the top contour of the image and similarly doing it for the image rotated upside-down. The cut positions are then derived as closely matching valleys of the original and the rotated images. Our proposed segmentation algorithm works well for different font styles, shapes and sizes better than the existing vertical projection profile based segmentation. The proposed algorithm has been tested on 1125 different word images, each containing multiple merged characters, from an old Kannada book and 89.6% correct segmentation is achieved and the character recognition accuracy of merged words is 91.2%. A few points of merge are still missed due to the absence of a matched valley due to the specific shapes of the particular characters meeting at the merges.

Recognition of open vocabulary, online handwritten pages in Tamil script

In this work, we describe a system, which recognises open vocabulary, isolated, online handwritten Tamil words and extend it to recognize a paragraph of writing. We explain in detail each step involved in the process: segmentation, preprocessing, feature extraction, classification and bigram-based post-processing. On our database of 45,000 handwritten words obtained through tablet PC, we have obtained symbol level accuracy of 78.5% and 85.3% without and with the usage of post-processing using symbol level language models, respectively. Word level accuracies for the same are 40.1% and 59.6%. A line and word level segmentation strategy is proposed, which gives promising results of 100% line segmentation and 98.1% word segmentation accuracies on our initial trials of 40 handwritten paragraphs. The two modules have been combined to obtain a full-fledged page recognition system for online handwritten Tamil data. To the knowledge of the authors, this is the first ever attempt on recognition of open vocabulary, online handwritten paragraphs in any Indian language.

Exploiting Chinese character models to improve speech recognition performance

The Chinese language is based on characters which are syllabic in nature. Since languages have syllabotactic rules which govern the construction of syllables and their allowed sequences, Chinese character sequence models can be used as a first level approximation of allowed syllable sequences. N-gram character sequence models were trained on 4.3 billion characters. Characters are used as a first level recognition unit with multiple pronunciations per character. For comparison the CU-HTK Mandarin word based system was used to recognize words which were then converted to character sequences. The character only system error rates for one best recognition were slightly worse than word based character recognition. However combining the two systems using log-linear combination gives better results than either system separately. An equally weighted combination gave consistent CER gains of 0.1-0.2% absolute over the word based standard system. Copyright © 2009 ISCA.

SPEECH RECOGNITION IN VODIS II.

VODIS II, a research system in which recognition is based on the conventional one-pass connected-word algorithm extended in two ways, is described. Syntactic constraints can now be applied directly via context-free-grammar rules, and the algorithm generates a lattice of candidate word matches rather than a single globally optimal sequence. This lattice is then processed by a chart parser and an intelligent dialogue controller to obtain the most plausible interpretations of the input. A key feature of the VODIS II architecture is that the concept of an abstract word model allows the system to be used with different pattern-matching technologies and hardware. The current system implements the word models on a real-time dynamic-time-warping recognizer.

Continuous speech recognition for the TIMIT database using neural networks

Four types of neural networks which have previously been established for speech recognition and tested on a small, seven-speaker, 100-sentence database are applied to the TIMIT database. The networks are a recurrent network phoneme recognizer, a modified Kanerva model morph recognizer, a compositional representation phoneme-to-word recognizer, and a modified Kanerva model morph-to-word recognizer. The major result is for the recurrent net, giving a phoneme recognition accuracy of 57% from the si and sx sentences. The Kanerva morph recognizer achieves 66.2% accuracy for a small subset of the sa and sx sentences. The results for the word recognizers are incomplete.