To assess sleep in the home situation, wrist actigraphy is often used. However, it requires an on-body sensor which may disturb sleep and primarily collects data on the movement of one wrist only. Video actigraphy, by estimating motion from captured infrared images, overcome these issues. In this paper, we compare activity levels from wrist and full body video actigraphy in a home setting. Video actigraphy correctly found 19 % more small and medium movements that were missed by the wrist sensor. We further show that similar values of sleep efficiency (SE, %) are obtained from simultaneous recordings of video and wrist actigraphy, and we compared both to reference SE values provided by a full polysomnography (PSG). The proposed video actigraphy proved convenient and easy to use in real home situations. It successfully found movements originating from under the blanket, and turned out robust to various sleeping positions, different illumination conditions, viewing angles, beds and blankets. Our results suggest that on-body actigraph sensors can be successfully replaced with the proposed video-based solution for the application of unobtrusive sleep monitoring and analysis.

The concept of eigen-MLLRs (Kuan-ting Chen et al., Proc. ICSLP 2000, p.742-5, 2000; Wang, N. J.-C. et al., Proc. ICASSP 2001, p.345-9, 2001), a variant of the eigenvoice method, is applied to unsupervised speaker enrollment in a large vocabulary CSR system. The emphasis is on fast adaptation. Two ways of estimating multiple eigen-MLLR transformations are introduced, either joint or separated with respect to the eigen-MLLR vector space. The first case allows multiple transforms to be robustly estimated from sparse data while the second achieves more accurate adaptation when more samples become available. The first decoded words spoken by a new test speaker are used to adapt the speaker-independent HMM means. The impact of this new enrollment algorithm is evaluated over a large real-life database dealing with professional medical transcriptions. Significant reductions of word-error-rates are achieved with less than 10 seconds of enrollment speech and without any supervision.

An algorithm for the generation of word graphs in a cross-word decoder that uses long span m-gram language models (LMs) is presented. The generation of word hypotheses within the graph relies on the word m-tuple-based boundary optimization. The graphs contain the full word history knowledge information since the graph structure reflects all LM constraints used during the search. This results in better word boundaries and in enhanced capabilities to prune the graphs. Furthermore, the memory costs for expanding the m-gram constrained word graphs to apply very long span LMs (e.g. ten-grams that are constructed by log linear LM combination) are considerably reduced. Experiments for lattice generation and rescoring have been carried out on the 5K-word WSJ task and the 64K-word NAB task.

The paper describes the improved handling of multiple pronunciations achieved in the Philips research decoder by (1) incorporating some prior information about their distributions and (2) combining the acoustic contributions of concurrent alternate word hypotheses. Starting from a baseline system where multiple pronunciations are treated as word copies without priors, an extension of the usual Viterbi decoding is presented which integrates unigram priors in a weighted sum of acoustic probabilities. Several approximations are discussed leading to new decoding aspects. Experimental results are presented for US broadcast news recordings. It is shown that the use of unigram priors has a clear positive impact on both error rate and decoding cost while the sum over multiple pronunciation contributions brings another small improvement. An overall 4% reduction of the error rate is achieved on the HUB-4 evaluation sets of 97 and 98.

This paper describes the new Philips Research decoder that performs large vocabulary continuous speech recognition in a single pass for cross-word acoustic models and an m-gram language model (with m up to 4) as opposed to our previous technique of multiple passes. The decoder is based on a time-synchronous beam search and a prex tree structure of the lexicon. Cross-word transitions are treated dynamically. A language-model look-ahead technique is applied on the bigram probabilities. On a variety of speech data, reduced error rates are obtained together with signi cant speed-ups con rming the advantage of an early use of all available knowledge sources. In particular, the search e ort of a one-pass trigram decoding is only marginally increased compared to bigram and the integration of cross-word triphones improves the overall accuracy by typically 10% relative.

This paper contains a description of the Philips/RWTH 1998 HUB4 system which has been build in a joint eort of Philips Research Laboratories Aachen and Aachen University o f T echnology.We will focus our discussion on recent improvements compared to the original 1997 HUB4 system and evaluate them on the HUB4'97 evaluation data.The paper will deal with 1. a rough system overview including feature extraction, acoustic training, audio stream segmentation, and decoding 2. log-linear interpolation of distance-language models, 3. and the integration of various acoustic and language models via Discriminative Model Combination (DMC).The performance of the described system is 23% (relative) better than the performance of the 1997 Philips HUB4 system.A w ord error rate of 17.9% was achieved on the 1997 HUB4 evaluation set, compared to 23.5% using the original 1997 system.

The paper deals with some aspects of production and perception of yes-no questions and non-final clauses in Dutch. The analysis of the prosodic characteristics of naturally produced utterances and the subjects’ responses obtained in the course a series of perception experiments with naturally produced and modified utterances yielded the following conclusions: • the identification of utterance type is to the largest extent determined by the presence of relevant prosodic cues in the nuclear accent • cues to utterance type in the pre-nuclear pattern appear to be used by listeners when the nuclear accent type allows of several possible interpretations • timing of the nuclear accent may be considered a relevant cue to utterance type, «early» vs. «late» rise accounting for «non-final» vs. «interrogative» preferences, «early» vs. «late» fall eliciting «statement» vs. «question/statement» reactions. INTRODUCTION The establishing of a set of categorically distinct intonation units forming the intonation system of a language is still one of the most challenging tasks for linguists. It is not yet clear how exactly the linguistic relevance of a certain type of intonation unit should be tested and some disagreement exists as to what differences should be called categorical rather than gradient [6]. There is no agreement either as to the function of intonation. Are intonation units associated with certain invariant categories of meaning and if so, in what terms should these meaningful distinctions be described? These and other fundamental questions are still open for discussion. At the same time it is evident that intonation is capable of changing the interpretation of an utterance and thus may be said to perform a certain communicative function, which finds its expression, among others, in the specific patterning of various utterance types. Previous research in the intonation of Dutch questions (e.g. [2], [3], [4]) has proved the already reported for other languages founding that the concept of interrogativity is mainly associated with a local terminal rise. Other cues to interrogativity include a raised register and absence of downtrend. Non-finality (or «continuation» we shall leave the terminological discussion beyond the scope of the present work) is also said to be cued by rising pitch. In her work devoted to the melodic marking of continuation versus questions in Dutch [2] J. Caspers claims that while the former are signalled by the accent-lending rise followed by sustained level pitch (1O in the IPO notation [1]), the latter are associated with a combination of an accent-lending rise and a final rise (12). One possible interpretation of these findings is that they provide extra evidence in support of the previously mentioned claim that questions are characterised by higher pitch values than other utterance types. The study reported here is concerned with finding further prosodic cues to non-finality and interrogativity. We limited our research exclusively to pitch characteristics (being fully aware of the communicative importance of other classes of prosodic features). The questions addressed were the following: • What are the most typical intonation patterns realised in non-final clauses and yes-no questions (in terms of nuclear and pre-nuclear patterns and their combinations) • To what extent can the perception of utterance type be influenced by the cues contained in the prenuclear vs. nuclear part of the utterance? • What is the role played by the timing of the nuclear accent in the perception of the utterance type ( Rise 1 vs. 2 and Fall A vs. (1)A in the IPO notation)? EXPERIMENTS 1: ANALYSIS AND PERCEPTION OF (FRAGMENTS OF) NATURALLY PRODUCED UTTERANCES 9 Yes-no questions with the inverted word order and 9 lexically and syntactically identical to them non-final clauses (mostly subordinate clauses of condition and concession also having the inverted word order) were embedded in appropriate contexts, printed on cards and read by 8 native speakers of Dutch (5 male, 3 female). The acquired recordings (144 utterances) were digitised and F0 contours were extracted using speech analysis programs WinCECIL (distributed by SIL as freeware) and PRAAT (developed by P. Boersma and D. Weenink). Pitch patterns produced by the speakers were transcribed using largely the notation developed in the framework of the Dutch School [1], with some modifications introduced in the course of the experimental analysis of the data, and then classified (following the British tradition in intonation analysis [7]) according to the type of nuclear and pre-nuclear pattern and frequency of occurrence in either type of utterance. 6 types of nuclear pitch accents were distinguished («&» means that the pitch movements are associated with one accented syllable, A* is used to indicate a delayed fall not distinguished in the IPO system): IPO Notation: Rise 1 (early ) 1 Rise 2 (late) (a)2, (b)A&2

The problem of unknown words has been addressed using automatically generated ller fragments which augment the lexicon and are incorporated in the language model. These fragments are used to reduce the damage on in-vocabulary words, to detect OOV regions and to provide a phonetic transcription for these regions. The performance of this technique has been evaluated in terms of damage reduction error rate and OOV tagging rate. Signi cant improvements are reported on both measures. In particular, the in uence of an appropriate tuning of the language model factor and word penalties is demonstrated as well as the usefulness of using cross-word triphones over fragments boundaries.

The combination of maximum likelihood linear regression (MLLR) with maximum a posteriori (MAP) adaptation has been investigated for both the enrollment of a new speaker as well as for the asymptotic recognition rate after several hours of dictation. We show that a least mean square approach to MLLR is quite effective in conjunction with phonetically derived regression classes. Results are presented for both ARPA read-speech test sets and real-life dictation. Significant improvements are reported. While MLLR achieves a faster adaptation rate when only few data is available, MAP has desirable asymptotic properties and the combination of both methods provides the best results. Both incremental and iterative batch modes are studied and compared to the performance of speaker-dependent training.

We address the problem of using word graphs (or lattices) for the integration of complex knowledge sources like long span language models or acoustic cross-word models, in large vocabulary continuous speech recognition. A method for efficiently constructing a word graph is reviewed and two ways of exploiting it are presented. By assuming the word pair approximation, a phrase level search is possible while in the other case a general graph decoder is set up. We show that the predecessor-word identity provided by a first bigram decoding might be used to constrain the word graph without impairing the next pass. This procedure has been applied to 64 k-word trigram decoding in conjunction with an incremental unsupervised speaker adaptation scheme. Experimental results are given for the North American Business corpus used in the November '94 evaluation.

We report on recent developments of the Philips large vocabulary speech recognition system and on our experiments with the Wall Street Journal (WSJ) corpus. A two-pass decoding has been devised that allows an easy integration of more complex language models. First, a word lattice is produced using a time synchronous beam search with a bigram language model. Next, a higher-order language model is applied to the lattice at the phrase level. The conditions insuring the validity of this approach are explained and practical results for trigram demonstrate its usefulness. The main system development stages on WSJ data are presented and our final recognizers are evaluated on Nov. '92 and Nov. '93 test-data for both 5 K and 20 K vocabularies.>

The paper presents a hybrid continuous-speech recognition system that leads to improved results on the speaker dependent DARPA Resource Management task. This hybrid system, called the combined system, is based on a combination of normalized neural network output scores with hidden Markov model (HMM) emission probabilities. The neural network is trained under mean square error and the HMM is trained under maximum likelihood estimation. In theory, whatever criterion may be used, the same word error rate should be reached if enough training data is available. As this is never the case, the idea of combining two different criteria, each of them extracting complementary characteristics of the feature is interesting. A state-of-the-art HMM system will be combined with a time delay neural network (TDNN) integrated in a Viterbi framework. A hierarchical TDNN structure is described that splits training into subtasks corresponding to subsets of phonemes. This structure makes training of TDNNs on large-vocabulary tasks manageable on workstations. It will be shown that the combined system, despite the low accuracy of the hierarchical TDNN, achieves a word error rate reduction of 15% with respect to our state-of-the-art HMM system. This reduction is obtained with a 10% increase only in the number of parameters.

This paper gives a status report of the Philips research system for phoneme-based, large-vocabulary, continuous-speech recognition. Like for many other systems, the recognition architecture is based on an integrated statistical approach. We describe the characteristic features of the system as opposed to other systems: 1. The Viterbi criterion is consistently applied both in training and testing. 2. Continuous mixture densities are used without tying or smoothing. 3. Time-synchronous beam search in connection with a phoneme look-ahead is applied to a tree-organized lexicon. The system has been successfully applied to the American English DARPA RM task. Here, we report experimental results for a German 13 000-word Philips internal dictation task. In addition to the scientific prototype, a PC version has been set up which is described here for the first time.

The author presents two fast anticipatory pruning schemes that have been investigated within the frame of a continuous speech recognition system based on HMM (hidden Markov modeling) and Viterbi decoding. Both algorithms rest on a coarse acoustic analysis in five broad phonetic categories. The first strategy interacts with the phoneme transitions during the search process, while the second takes account of lexical constraints by operating at the word level. Experiments have been performed on a thousand-word continuous speech task with no language model. Results show that a significant computational reduction is feasible without impairing the overall accuracy.>