Showing papers on "Word error rate published in 1976"
••
TL;DR: The findings were interpreted as showing that more than one type of coding process can underly the word superiority effect in word recognition when phonetic information does not discriminate well among response alternatives.
Abstract: An experiment was designed to examine the contribution of phonetic information in the processing of words in tachistoscopic recognition masking. Following stimulus presentation, subjects were required to indicate which of two alternatives had appeared. On trials containing word stimuli, the alternatives were either phonetically identical (SENT, CENT) or not (SOLD, COLD). Recognition performance was inferior in the former case, provided conditions were not structured to discourage reliance on phonetic information. The findings were interpreted as showing that more than one type of coding process can underly the word superiority effect. Phonetic information is ordinarily used to code words in this type of task, but an alternative processing tactic (e.g., one relying on visual or perhaps semantic codes) can also be effectively used in word recognition when phonetic information does not discriminate well among response alternatives.
125 citations
••
TL;DR: In this paper, statistical techniques are described which are useful for simulating tails of distributions and the importance-sampling procedure is used to modify the probability densities of the input values in a way that makes simulation possible.
Abstract: Simulation of repeaters for an optical-fiberguide digital-communication system requires the calculation of the statistical error rate of the signal. The calculations of these error rates are difficult because of the non-gaussian nature of the noise in the optical detector. In this paper, statistical techniques are described which are useful for simulating tails of distributions. In particular, the importance-sampling procedure is used to modify the probability densities of the input values in a way that makes simulation possible. Application of this procedure gives more accurate results in reasonable computer times. The method is applied to the calculation of the error rate of a fiberguide repeater. Realistic examples are simulated. Results compare favorably with experimental measurements. The number of samples needed for simulation was reduced by five to six orders of magnitude. An alternative numerical solution to the problem is also developed.
74 citations
••
TL;DR: Examination of several techniques for integrating an independent contextual postprocessor (CPP) into a full classification system finds that a standardized CPP can be built independently of the rest of the classification system.
50 citations
••
TL;DR: In this article, the bias of the apparent error rate is derived in the context of two multivariate normal populations with unknown different means and unknown common covariance matrix and a correction term is available for reducing the bias from the first to second order with respect to the reciprocals of the initial sample sizes.
Abstract: SUMMARY The apparent error rate is a commonly used estimator of the actual error rate in discriminant analysis. In this study the asymptotic bias of the apparent error rate is derived in the context of two multivariate normal populations with unknown different means and unknown common covariance matrix. From the derived expansion a correction term is available for reducing the bias of the apparent error rate from the first to the second order with respect to the reciprocals of the initial sample sizes. Also, some previously unanswered questions on inequalities between the average apparent, the optimal, and the average actual error rates are solved.
48 citations
••
TL;DR: Results indicated no effect for letter difficulty, although recognition latency reliably decreased with word frequency and monotonically increased with word length, suggesting that the authors do not read letter by letter, but that whatever plays a role in word recognition is smaller than the word and correlated withword length in letters.
Abstract: In a paradigm that avoids methodological problems of earlier studies, evidence was gathered addressed to the question of whether we read letter by letter. If word recognition involves letter recognition, then the difficulty of recognizing a word should vary with the difficulty of recognizing its letters. This was tested by assessing letter difficulty in two letter discrimination tasks and in a letter naming task, and then comparing 15 adult subjects’ visual recognition latency to 72 easy-letter words and to 72 difficult-letter words. Word frequency and word length were also manipulated. Results indicated no effect for letter difficulty, although recognition latency reliably decreased with word frequency and monotonically increased with word length (21 msec/letter), suggesting that we do not read letter by letter, but that whatever plays a role in word recognition is smaller than the word and correlated with word length in letters.
37 citations
•
12 Jul 1976TL;DR: In this paper, a method of and an apparatus for determining the best match between a search word and a plurality of file words is disclosed, which involves simultaneously comparing the search word to a file word and to a match word.
Abstract: A method of and an apparatus for determining the best match, i.e., minimum Hamming distance, between a search word and a plurality of file words is disclosed. The method involves simultaneously comparing the search word to a file word and to a match word. If the comparison indicates that the file word is a better match than the match word, the file word becomes the match word and the next file word is compared to the new match word. The operation continues until all the file words have been operated upon such that the last match word becomes the best match word.
23 citations
••
01 Apr 1976TL;DR: This paper describes a connected speech understanding system being implemented in Nancy made up of an acoustic recognizer which gives a string of phoneme-like segments from a spoken sentence, a syntactic parser which controls the recognition process, a word recognizer working on words predicted by the parser and a dialog procedure which takes in account semantic constraints in order to avoid some of the errors and ambiguities.
Abstract: This paper describes a connected speech understanding system being implemented in Nancy, thanks to the work done in automatic speech recognition since 1968. This system is made up of four parts : an acoustic recognizer which gives a string of phoneme-like segments from a spoken sentence, a syntactic parser which controls the recognition process, a word recognizer working on words predicted by the parser and a dialog procedure which takes in account semantic constraints in order to avoid some of the errors and ambiguities. Some original features of the system are pointed out : modularily (e.g. the language used is considered as a parameter), possibility of processing slightly syntactically incorrect sentences, ... The application both in data management and in oral control of a telephone center has given very promising results. Work is in progress for generalizing our model : extension of the vocabulary and of the grammar, multi-speaker operation, etc.
16 citations
••
TL;DR: The performance of the verification component has been measured by obtaining the distribution of verification scores for all word hypotheses generated by the speech understanding system, and determining the scores for words that should...
Abstract: Given a word whose presence has been hypothesized in an unknown utterance, one way of enhancing the confidence in that hypothesis is to generate a synthetic parameterization of the word and then match it against the equivalent parametric representation of the unknown utterance We have implemented such an approach in the speech understanding system under development at Bolt Beranek and Newman, Inc Given a word, a synthesis‐by‐rule program generates a representation in terms of linear prediction spectra, which are matched against similar spectra of the raw signal using a 13‐pole linear prediction error metric in conjunction with a dynamic programming time‐normalization algorithm Some automatic talker normalization procedures have been implemented in the synthesis strategy The performance of the verification component has been measured by obtaining the distribution of verification scores for all word hypotheses generated by the speech understanding system, and determining the scores for words that should be verified correctly versus those scores for false word hypotheses [Supported by ARPA under Contract No N00014‐75‐C‐0053]
15 citations
••
01 Apr 1976TL;DR: Verification offers an alternative strategy by doing a top-down parametric word match independent of segmentation and labeling, which results in a distance measure between the reference parameterization of a hypothesized word and the computed parameterizations of the real speech.
Abstract: If, in a speech understanding system, word matching is performed at the phonetic level, then the accurate determination of the locations and identities of words present in an unknown utterance is necessarily limited by the phonetic segmentation and labeling. Verification offers an alternative strategy by doing a top-down parametric word match independent of segmentation and labeling. The result is a distance measure between the reference parameterization of a hypothesized word and the computed parameterization of the real speech. This distance is interpreted as the likelihood of that word having actually occurred over a given portion of the utterance.
13 citations
••
12 Apr 1976TL;DR: The voice-operated question-answering system for seat reservation is constructed by computer simulation technique and the promising results are obtained.
Abstract: The speech recognition system composing a part of the question-answering system operated by conversational speech is described. The recognition system consists of two stages of process : acoustic processing stage and linguistic processing stage. In the acoustic processing stage, input speech is analyzed and transformed into the phoneme sequence which usually contains ambiguities and errors caused in the segmentation and phoneme recognition. In the linguistic processing stage, the phoneme sequence containing ambiguities and errors is converted into the correct word sequence by the use of the linguistic knowledge such as phoneme rewriting rules, lexicon, syntax, semantics and pragmatics. The voice-operated question-answering system for seat reservation is constructed by computer simulation technique and the promising results are obtained.
11 citations
••
TL;DR: Experiments on elongated and shortened words also show that perceived length directly influences word responses, indicating that word length operates as a separate cue in the recognition process.
••
12 Apr 1976
TL;DR: The Speech Communications Group at SPERRY UNIVAC Defense Systems is developing a linguistically-oriented procedure for recognizing words, phrases, and natural sentences by computer that will be extended from the recognition of several-word noun phrases to the understanding of more natural sentences.
Abstract: The Speech Communications Group at SPERRY UNIVAC Defense Systems is developing a linguistically-oriented procedure for recognizing words, phrases, and natural sentences by computer. The major components of the current speech recognition system perform acoustic and phonetic analysis, phonetic segmentation, and lexical matching and scoring. The acoustic processing is based on a linear-predictive spectral analysis of the speech signal. Sounds are classified by manner, place, and voicing using formant frequencies and other spectral functions, as well as information about syllable boundaries and nuclei. A linear sequence of analysis segments is created, and matched against the lexicon using a scoring matrix that ranks analysis-lexical segment pairs by their expected confusions. Word sequences are progressively formed and ranked against the entire input to determine the most likely phrases spoken. When the recognition system was tested on a 31-word vocabulary from two male speakers, single word recognition scores of 95% correct were obtained when the task syntax was used. Preliminary results for recognizing connected word sequences from three male speakers range from 54 to 74% for a task with constrained word order. Current plans for enhancing the recognition system include the incorporation of components for phonological rules, speaker normalization, and prosodic guidelines. By adding more powerful procedures for syntactic and semantic analysis, the system will be extended from the recognition of several-word noun phrases to the understanding of more natural sentences.
••
01 Apr 1976TL;DR: An online automatic spoken word recognition system has been developed for the researches on the automatic recognition of speech and the possible combination of phonemes to every word was found to be 94% when the speakers were permitted to repeat their utterances for three times.
Abstract: An online automatic spoken word recognition system has been developed for the researches on the automatic recognition of speech. In this system, the spoken word is first frequency analysed with a filter bank of single tuned low selectivity filters. Three major local peaks in the spectrum and the amplitude of the speech wave are extracted every 10 ms. The frequencies of two local peaks are used for classifying the vowels, and the frequencies of three local peaks, the movements of them and the amplitude are used for classifying the semi-vowels and consonants. Input speech is thus transformed into a sequence of the notations expressing the phonemes or phoneme groups every 10 ms. The sequence is again transformed into possible phonemic strings which are called input words henceforth and are convenient for the comparison with the contents of the dictionary. The Hamming's distance between each input word and each item of the contents of the dictionary is computed where the notations of phonemes and phoneme groups are expressed by 9 bits binary vectors. The item in the dictionary nearest to one of the input word is selected as the output of the recognition system. The experiments were carried out with the utterance of five speakers from whose utterances the standard patterns for P1, P2 and Pe3 distribution had been made. The recognition score was 96% for the 20 city names involving all kinds of phonemes. The speech samples were increased to 166 city names and 82% of the utterances of three speakers were correctly recognized by adding the possible combination of phonemes to every word. Next, 13 different speakers uttered 51 city names having long distance between each other, the recognition score was found to be 94% when the speakers were permitted to repeat their utterances for three times.
••
01 Aug 1976TL;DR: For random errors the optimum data-block length is found to be proportional to √(a/k), and the maximum efficiency a function of ka, where a is the message overhead and k is the error rate.
Abstract: The study considers the effect of error rate, block length, transmission delay, and error-recovery mechanism on the efficiency of the communications channel in computer-computer communications. The systems investigated are full-duplex communications with unlimited and limited buffering, and half-duplex communications. Both random and burst errors are considered. For all cases, over a significant range of parameters, closed-form expressions for the optimum block length, and the corresponding efficiency are obtained. For random errors the optimum data-block length is found to be proportional to √(a/k), and the maximum efficiency a function of ka, where ais the message overhead and k is the error rate. The burst error results are similar.
••
TL;DR: A scheme suitable for selection of features for OCR systems has been developed and successfully applied to several type fonts resulting in systems which recognize upper and lower case alphanumerics with less than one error per 10 000 processed characters.
•
19 Apr 1976
TL;DR: In this paper, the error rate of the phase difference between the transmission wave and the reflective wave was reduced by reducing the error of measuring distance against the object, by using a phase difference measure.
Abstract: PURPOSE:To reduce the error of measuring distance against the object, by reducing the error rate of the phase difference between the transmission wave and the reflective wave.
••
TL;DR: In this paper, a definition of error rate is proposed in which the number of significant outcomes needed to answer the question of interest is considered and a distinction is made between tests of fundamental importance and those of only subsidiary interest.
Abstract: No current method of controlling error rate is appropriate for all experiments. When the error rate is set at traditional levels a per comparison error rate can yield too high a proportion of Type I errors, while an experimentwise error rate can be too conservative because the purpose of the experiment is not taken into account. A definition of error rate is proposed in which the number of significant outcomes needed to answer the question of interest is considered and a distinction is made between tests of fundamental importance and those of only subsidiary interest. The definition provides a systematic method of unequally allotting the error rate such that more power is provided for tests of crucial interest and for experiments in which several significant results are required.
••
01 Apr 1976TL;DR: Initial word reference signals, built from the finite set of the reference elements and written by means of acoustic-phonetic transcription, and reference phrase signals, which are composed of reference word signals with the help of special rules are introduced.
Abstract: Initial word reference signals are built from the finite set of the reference elements and written by means of acoustic-phonetic transcription. The transformations of the initial word references permitting to obtain reference signals of words with the different rate of pronunciation are introduced. Reference phrase signals are composed of reference word signals with the help of special rules. The recognition of a given signal consists in location of the reference signal of a maximum likelihood for the given one and in indication of a word or a word sequence which composes it. Learning algorithm, which finds the transcriptions of words and the set of reference elements allows to adapt to some announcer and to change some part of the dictionary or the whole one. Described method of a recognition gives 1% of errors, 3% of refusals for 300 words. The word recognition in connected speech gives 3% of errors and 7% of refusals.
••
CERN1
TL;DR: A signal detection method for delta-coded speech is presented which exploits the redundancy of the source to reduce the error rate owing to additive Gaussian noise in the channel.
Abstract: A signal detection method for delta-coded speech is presented which exploits the redundancy of the source to reduce the error rate owing to additive Gaussian noise in the channel. The method implements a Bayes strategy with dual decision boundaries selected by a 6th-order digital predictor. Fixed decision boundary pairs, appropriate for a wide range of input signal/noise ratio, are established. This easily implemented technique is shown to give an error-rate reduction of about 40% in coherent antipodal and coherent and noncoherent orthogonal communication.
••
TL;DR: The digital part of a simplex binary telemetry system for conveying control signals to some remote body such as a missile is described and the effect of asymmetry in the binary channel is examined.
Abstract: This paper describes the digital part of a simplex binary telemetry system for conveying control signals to some remote body such as a missile. The decoder in the remote body is small, light, has a low power consumption, and is designed to operate above a certain noise threshold. When the signal-to-noise ratio of the signal at the input to the decoder is low the latter only accepts information which has a high probability of being correct thereby causing the rejection of most of the transmitted information. When information is rejected the remote body functions on the data which were last accepted by the decoder.Both experimental and theoretical results are presented for error rate and received frame rate, and the effect of asymmetry in the binary channel is examined.