Consonant Part Research Articles

The Influence of Shandong dialect on English learning

Phonologythe material shell of languagehas its own peculiarities because language structures and realizations vary. Behaviorist phonological theory implies that prior phonological patterns interfere with second-language communication. Learners inadvertently transfer their native language pronunciation to foreign language acquisition. The transfer is a psychological phrase that used to refer primarily to the influence of former learning on subsequent learning but now refers to the influence of one type of learning on another. Positive and negative transfers exist. English is Indo-European, while Chinese is Sino-Tibetan. distinct language families mean distinct phonological intonation systems and pronunciation patterns, therefore negative transfer of phonological learning is inevitable. Chinese students learn various dialects, each with its own system, and these dialects deeply affect English phonological intonation learning. This paper examines the transfer effect of the Shandong dialect on English phonology and the negative transfer of dialects to English intonation learning. The phonological system differs in terms of consonant parts and pronunciation methods, and the intonation system differs in tone, intonation, stress, and rhythm.

Peculariarities of neologisms in French Internet slang

Our paper deals with neologisms in the French language introduced through the Internet. In the modern world, Internet slang plays important role in the communication of Internet users; Recently, there has been a tendency to actively use Internet vocabulary in the daily life of French youth.  The French language is characterized by shotering a large number of words in colloqual speech. This phenomenon also manifests itself in Internet relationships. Distinctive features include active use of abbreviations, rep;acement of some letter combinations with apostrophes, use of sumbols (numbers, capital letters, money symbols) to replace consonant parts of whole words, creation and/or use of neologisms. French Internet slang, which is characterized by different ways of word formation, often atypical for the literary French language, is widely used by modern French young people when communicating in Internet sefvices, and it comes out of the framework of the virtual language, even in resal everyday communication.

Lyrics-to-Audio Alignment by Unsupervised Discovery of Repetitive Patterns in Vowel Acoustics

Most of the previous approaches to lyrics-to-audio alignment used a pre-developed automatic speech recognition (ASR) system that innately suffered from several difficulties to adapt the speech model to individual singers. A significant aspect missing in previous works is the self-learnability of repetitive vowel patterns in the singing voice, where the vowel part used is more consistent than the consonant part. Based on this, our system first learns a discriminative subspace of vowel sequences, based on weighted symmetric non-negative matrix factorization (WS-NMF), by taking the self-similarity of a standard acoustic feature as an input. Then, we make use of canonical time warping (CTW), derived from a recent computer vision technique, to find an optimal spatiotemporal transformation between the text and the acoustic sequences. Experiments with Korean and English data sets showed that deploying this method after a pre-developed, unsupervised, singing source separation achieved more promising results than other state-of-the-art unsupervised approaches and an existing ASR-based system.

Non-Touch Character Input System Based on Hand Tapping Gestures Using Kinect Sensor

There have been a lot of studies on the text input system using the image-based hand gesture recognition. However, hand gesture languages such as sign languages, finger alphabets, and aerial handwriting treated in the previous works have some problems to be commonly used. The aerial handwriting requires much time for writing and recognition. The sign languages and finger alphabets demand quite a knowledge and practice for using it, which results in restricting the number of their users. As a solution to the problems, this paper proposes a new character input system based on hand tapping gestures for Japanese hiragana and English characters that can be used to facilitate human–computer interaction. The hand tapping gestures are motions for tapping keys on aerial virtual keypads by hands, which can be effectively used as a hand alphabet by anyone, including hearing impaired individuals. For hiragana characters, the hand used for tapping a key and the number of stretched fingers of the hand decide the consonant part of characters, and thereby the aerial virtual keypad. The character to be entered is determined by tapping the key on the virtual keypad corresponding to the desired vowel. Because we adopt a key layout similar to the Japanese and English flick keyboard of smart phones, our hand tapping gestures can be easily used by anyone with only a brief description. The users can effectively interact with computers by using our non-touch input system where only the Kinect sensor is used without any keyboard, mouse or body-worn device. We expect that our character input system will open a new channel for human-computer interaction.

Separation and integration of sound sources in auditory processing

Some perceptual experiments were conducted to elucidate sound source separation processing in speech perception. The phenomenon of phonemic restoration was tested if it occurs in dichotic listening. An original speech is Japanese /VCV/, the stimulus is made by manipulating the /VCV/ as was replaced the consonant part with noise. The duration of replacement was varied at six levels: 10, 20, 40, 60, 80, and 100 ms. There were two hearing modes: in one case, the monaural stimulus was presented to both the ear, and in the other case, the speech part of the stimulus was presented to an ear and the noise part to the opposite ear simultaneously. Thirty subjects in the age of twenties participate the listening test. The results show that phonemic restoration effect is clearly observed in both hearing modes. The rate of restoration is 92%, and 95% even in dichotic listening. There was no feature of speech in the noise has no relation with speech sound. Moreover, the direction of the noise sound was opposite to that of speech sound source, different in 180 degree. Even though the two sound signal are somehow integrated to hear restored speech.

Stop Consonant-Short Vowel (SCSV) Classification for Tamil Speech Utterances

Tamil Language is one of the ancient Dravidian languages spoken in south India. Most of the Indian languages are syllabic in nature and syllables are in the form of Consonant-Vowel (CV) units. In Tamil language, CV pattern occurs in the beginning, middle and end of a word. In this work, CV units formed with Stop Consonant – Short Vowel (SCSV) were considered for classification task. The work carried out in three stages, Vowel Onset Point (VOP) detection, CV segmentation and classification. VOP is an event at which the consonant part ends and vowel part begins. VOPs are identified using linear Prediction residuals which provide significant characteristics of the excitation source. To segment the CV units, fixed length spectral frames before and after VOPs are considered. Both production based features - Linear Predictive Cepstral Coefficients (LPCC) and perception based features - Mel Frequency Cepstral Coefficients (MFCC) are extracted and given as input to the classifiers designed with multilayer perceptron and support vector machine. A speech corpus of 200 Tamil words uttered by 15 native speakers was used, which covers all SCSV units formed with Tamil stop consonants (/k/,/ch/,/d/,/t/,/p/) and short vowels (/a/,/i/, /u/, /e/, /o/). The classifiers are trained and tested for its performance using various measures. The results indicate that the model built with MFCC using support vector machine RBF kernel outperforms.

RECOGNITION OF TAMIL SYLLABLES USING VOWEL ONSET POINTS WITH PRODUCTION, PERCEPTION BASED FEATURES

Tamil Language is one of the ancient Dravidian languages spoken in south India. Most of the Indian languages are syllabic in nature and syllables are in the form of Consonant-Vowel (CV) units. In Tamil language, CV pattern occurs in the beginning, middle and end of a word. In this work, CV Units formed with Stop Consonant - Short Vowel (SCSV) were considered for classification task. The work carried out in three stages, Vowel Onset Point (VOP) detection, CV segmentation and classification. VOP is an event at which the consonant part ends and vowel part begins. VOPs are identified using linear prediction residuals which provide significant characteristics of the excitation source. To segment the CV units, fixed length spectral frames before and after VOPs are considered. In this work, production based features, Linear Predictive Cepstral Coefficients (LPCC) and perception based features, Perceptual Linear Predictive Cepstral Coefficients (PLP) and Mel Frequency Cepstral Coefficients (MFCC) are extracted which are used to build the SCSV classifier using multilayer perceptron and support vector machine. A speech corpus of 200 Tamil words uttered by 15 native speakers was used, which covers all SCSV units formed with Tamil stop consonants (/k/, /ch/, /d/, /t/, /p/) and short vowels (/a/, /i/, /u/, /e/, /o/). The classifiers are trained and tested for its performance using predictive accuracy measure. The results indicate that perception based features, MFCC and PLP provides better results than production based features, LPCC and the model built using support vector machine outperforms.

Influence of duration on the perception of consonants /x/ and /j/ in Chinese

In the current study, we investigated whether Mandarin Chinese native speakers' perception of /x/ and /j/ was affected by the duration of consonant parts. Two perceptual experiments, in which the method of constant stimuli was employed, were conducted. Six normal-hearing adults (four females) took part in the experiment. Their average age was 24 yr. All subjects were native speakers of Mandarin Chinese. In the first experiment, Chinese syllables that begin with /x/ or /j/ were extracted from a speech database, and the consonant parts were manipulated in terms of duration. As the duration of /x/ was decreased or the duration of /j/ was increased, to a certain extent, the consonant which had been originally /x/ was perceived as /j/, and vice versa. Synthesized noises instead of recorded consonants were utilized in the second experiment, similar effects of the consonant duration appeared.

Influence of the major connector in a maxillary denture on phonetic function

The purpose of this study was to investigate the influence of the major connector in a maxillary denture. Eight patients with sound dentition were recruited in this study. The experimental connectors were fabricated with Co-Cr. Two positional designs were used: one was the middle type (M), and the other was the anterior and posterior type (AP). The six sounds, /shi/, /hi/, /chi/, /ki/, /ri/, and /ni/, were chosen as the test sounds. Using a speech recognition system, the intelligibility, change and duration of the test sounds were evaluated under three conditions: without connector (WO·C), with the M-type connector (W·M), and with the AP-type connector (W·AP). The intelligibility in WO·C and W·AP was 86.9 ± 17.4% and 57.1 ± 30.1% for /shi/, 79.2 ± 19.7% and 70.5 ± 27.0% for /hi/, 74.7 ± 14.6% and 43.6 ± 24.3% for /chi/, 54.4 ± 16.5% and 23.2 ± 17.8% for /ki/, 85.2 ± 15.9% and 55.1 ± 30.6% for /ri/, and 76.9 ± 26.9% and 50.2 ± 31.3% for /ni/. Significant differences were recognized between WO·C and W·AP in /shi/, /chi/, /ki/ and /ri/(P<0.01). The intelligibility in W·M was higher than that in W·AP in all test sounds. As for incorrect labels, significant differences were recognized in the appearance ratio of /hi/ as the incorrect label in the test sound /shi/ (P<0.05). There were significant differences in some incorrect labels of /hi/ and /chi/. As for duration of consonant part, there was no significant difference in each test sound. It was indicated that the middle type of the major connector would be recommended from a viewpoint of phonetic function.

Contribution of Spectrotemporal Features on Auditory Event-Related Potentials Elicited by Consonant-Vowel Syllables

To determine the influence of spectrotemporal properties of naturally produced consonant-vowel syllables to speech-evoked auditory event-related potentials (ERPs) for stimuli with very similar or even identical wide-band envelopes. Speech-evoked ERPs may be useful for validating the neural representation of speech. Speech-evoked ERPs were obtained from 10 normal-hearing young adults in response to the syllables /da/ and /ta/. Both monosyllables were obtained from ongoing speech. They have quite similar wide-band envelopes, and they mainly differ in the spectrotemporal content of the consonant parts. Additionally, each stimulus derivatives were investigated with (1) isolated consonant part ("consonant stimulus"), (2) isolated vowel part ("vowel stimulus"), and (3) removed spectral information but identical wide-band envelope. Latencies and amplitudes of the N1 and P2 components were determined and analyzed. ERPs in response to the naturally produced /ta/ syllable had significant shorter N1 and P2 latencies and larger amplitudes than ERPs in response to /da/. Similar differences were observed for the ERPs evoked by the consonant stimuli alone. For the vowel stimuli and stimuli with removed spectral information, no significant differences were observed. In summary, differences between the ERPs of /da/ and /ta/ corresponded to the distinct spectrotemporal content in the consonant parts of the original consonant-vowel (CV) syllables. The study shows that even small differences in spectrotemporal features of speech may evoke different ERPs, despite very similar or even identical wide-band envelopes. The results are consistent with a model that ERPs evoked by short CVs are an onset response to the consonant merged with an acoustic change complex evoked by the vowel part. However, all components appear as one P1-N1-P2 complex. The results may be explained by differences in the narrow-band envelopes of the stimuli. Therefore, this study underlines the limitations of the wide-band envelope in explaining speech-evoked ERPs. Additionally, the results of this study are of special interest for clinical application since some of the ERP parameter differences, as the N1 latency, are present not only in the ERPs of each single subject but also in the group mean value of all N1 latencies. Thus, presented ERP measurements in response to CVs might be used for identification of potential problems in phoneme differentiation caused by spectrotemporal analysis problems.

The minimum word in Warray*

Words in Warray are minimally bimoraic, a constraint on minimum word size whose effects are evident in most languages of the world. We examine the differing satisfactions of this constraint for nouns and verbs in Warray from the perspectives of Optimality Theory. Nouns of the underlying shape /CVC/ appear with a long vowel [CVVC]. Verbs with the underlying shape /CVC/ appear without a long vowel [CVC]. We propose that this systematic difference is not categorially stipulated, but rather that it follows from another systematic difference between nouns and verbs. In nouns the coda consonant forms part of the root, whereas in verbs it is a separate suffix. We propose that final consonants are generally not morale in Warray, and consequently the nouns appear as [CVVC]. However with verbs, the general requirement that root‐level suffixes should align with the root prevents vowel lengthening, and requires that the final consonant should be moraic.

Encoding method for syllables

A method for encoding syllables of a language, particularly the Japanese language, and for facilitating the extraction of sound codes from the input syllables, for voice recognition or voice synthesis includes the step of providing a syllable classifying table, in which each syllable is represented by an upper byte code indicating the consonant part of the syllable and a lower byte code indicating the non-consonant part of the syllable. The consonants constitute a first category of data classified by phonetic features, while the non-consonants constitute a second category of data classified by phonetic features, so that the extraction of consonant or non-consonant sounds can be made by a search in only the first or the second categories. The encoding of diphthongs are made in such a manner that those containing the same vowel have the same remainder corresponding to the code of this vowel, when the codes are divided by the number of vowels contained in the second category, so that the extraction of a vowel from diphthongs can be achieved by a simple mathematical division.

Segmentation and recognition of handwritten pitman shorthand outlines using an interactive heuristic search

A new approach to the overall segmentation and interpretation of the consonant part of shorthand outlines is proposed in this paper. In the approach an interactive heuristic search scheme allows communication between bottom-up and top-down recognition processes. The communication, in turn, leads the direction of search towards the best or nearly the best interpretation of the consonant outline. Experiments using a set of 2058 shorthand outlines from five writers show that, in comparison with a non-interactive technique, the approach is efficient and produces an improvement from 65 to 85% in correct outline segmentation and an improvement from 48 to 57% in correct recognition of complicated consonant outlines.

Using auditory models to enhance telephone speech for the hearing impaired

Previously the effectiveness of frequency shaping and compression achieved via manipulation of the short-term spectrum of telephone speech for mild-to-moderate hearing impaired individuals was presented [Terry etal., J. Acoust. Soc. Am. 88, S169 (1990)]. This paper compares this technique with two analysis/resynthesis techniques designed to enhance the consonant parts of speech. In one system a time-domain auditory model (similar to the model proposed Goldstein [Hear. Res. 49, 39–60 (1990)] was used to analyze the speech. A set of consonant identifying features was then used to drive the resynthesis of the speech from the periodicity spectrum. In the other system the speech was resynthesized from LPC coefficients obtained via LPC analysis of the telephone speech signal. Again a set of consonant identifying features derived from LPC analysis was used to drive the resynthesis. The consonant-enhanced telephone band speech was presented to subjects with sensori-neural hearing loss. A consonant-confusion analysis will be presented comparing and evaluating both analysis/resynthesis techniques with a frequency shaping technique based on the subjects audiogram. [Work supported by US West.]

Special speech recognition approaches for the highly confusing Mandarin syllables based on hidden Markov models

In this paper, several special speech recognition approaches based on hidden Markov models (HMMs) are presented for the highly confusing Mandarin syllables by considering the characteristics of the vocabulary. This is because there are totally 408 syllables (disregarding the tones) in Mandarin speech, and it is believed that correct recognition of these syllables is the key to the development of a Mandarin dictation machine which recognizes Mandarin speech with very large vocabulary and unlimited texts. However, accurate recognition of these 408 syllables is very difficult because there exist 38 confusing sets among them, each of which has at most 19 very confusing syllables. Direct application of conventional standard approaches of HMMs to these syllables gives recognition rates in the order of only 70–80%, thus several special approaches are proposed in this paper to provide better performance. Some of these approaches concentrate on the training algorithms for the HMMs, including the two-pass training, the revised two-pass training, the three-pass training, and the revised three-pass training approaches. The basic idea is to protrude the very short initial parts (initial consonant parts) and de-emphasize the final parts (vowel or diphthong parts but including possible medial or nasal ending) of the syllables such that the confusing syllables can be better distinguished. Also, some other approaches concentrate on the recognition phase of the HMMs, including the state duration bounds and the two-stage search strategy, which can also improve the recognition rates and/or speeds. A special hardware is also implemented to complete all the recognition operations in real-time, on which all the approaches discussed in this paper can be applied.

デジタル音声処理の補綴学的応用

The purpose of this study was to investigate the acoustic characteristics of the nine sounds, /sa/, /si/, /ka/, /ki/, /ci/, /hi/ and /ri/, which were reported affected by wearing complete denture, through the acoustic analysis of speech samples uttered by nine male subjects. Spectrum analysis was done within the frequency range of 0 to 10 KHz according to the auto-correlation method of liner predictive coding (LPC) analysis. Results 1. The consonant parts of /sa/, /si/, /ci/ and /hi/ wee characterized by the spectrum envelopes and their peak frequencies derived from the LPC analysis of order 5. 2. The consonant parts of /ka/ and /ki/ were characterized by their first spectrum peak frequencies derived from the LPC analysis of order 10. 3. The consonant part of /ri/ was characterized by the spectrum envelope derived from the LPC analysis of order 24, and the sound waveform. 4. The frequencies of the formants (F 1, F 2, F 3) of the following vowels /i/ were derived from the LPC analysis of order 24. They were influenced by individual variations and preceding consonants. 5. Each sound had its own characteristics in the duration of the consonant and of the closure interval and the time from the end of preceding vowel to the beginning of the following vowel.

人工口蓋床の形態が発音に及ぼす影響に関する研究

The purpose of this study was to investigate the influence of the alteration of palatal form on pronunciation. Three types of artificial palatal plates simulating the upper complete denture, which artificial posterior teeth position was more lingually than natural dentition, were made for each dentulous subject. Difficulty of articulatory movements and sound qualities were assessed when the subjects pronounced /sa/, /si/, /ka/, /ki/, /ci/, /hi/, and /ri/ with each plate. The distortional sounds judged by auditory psychological experiment were analysed by acoustic methods, and mandibular movements during pronunciation of them were measured. 1. /ki/ and /hi/ were affected remarkably when the posterior position was more lingually, and /ri/, /ci/ and /sa/ were also affected. 2. The acoustic characteristics of distortional sounds were estimated by peak trajectories, supectrum envelopes, peak frequencies and sound waveforms of consonant parts, formant frequencies and formant trajectories of following vowels, and segemental durations. 3. Mandibular movements did not influence directory on the acoustic characteristics of distortional sounds. It was thought that tongue and lips mainly influenced on those.

Nonlinear time‐scale modification of speech signal with varied segmental duration characteristics

Segmental duration of each phoneme changes depending upon the speaking rate. Generally, vowel parts are easier to be compressed or expanded than consonant parts are in fast or slow speech, respectively. Questions raised in this paper include how the speaking rate can be extracted from the speech signal without knowing the content (i.e., phonetic information) and what kind of time‐scale modification can be chosen in order to control speaking rate. First, the segmental duration compressibility of the speech signal was defined by path slopes in DTW spectral matching when utterances with various kinds of speaking rates were matched to a reference utterance of a normal speaking rate. On the assumption that the compressibility is inversely proportional to segmental spectrum changes, the relationship between the compressibility and the average cepstral time difference Δcep [S. Furui, IEEE Trans. Acoust. Speech Signal Process. ASSP‐34, 52–59 (1986)] was studied. The results showed that the Δcep is an efficient parameter to represent the compressibility. By using the Δcep as a control parameter of the speaking rate, nonlinear time‐scale modification can be achieved without speech quality degradation.

Hearing aid using an emphasized speech for hearing‐impaired subjects

A new hearing aid which emphasizes the phonetic features of monosyllables is suggested in this paper. In vowel parts of speech, we applied a digital filtering technique to a formant emphasis. The 101 order transversal filter emphasizes formant frequencies (+ 10 − + 30 dB). This filter enabled us to reduce misrecognitions of /i/ and /e/, /u/ and /o/ for hearing‐impaired children. To emphasize consonant parts, three emphasis methods are experimented and discussed. One is an emphasis of amplitudes in consonant part, the next is a duration expansion of nasal consonant, and the last is an insertion of silence between stop consonants and the following vowels. The combinations of these methods were also experimented. The results of these experiments showed a distinct improvement in the recognition ratio of monosyllables.

東京市立聾學校に於ける發音調査の報告

This is a report of the articulation test applied to 166 deaf boys and girls of our school (from 3rd year grade of elementary course to 4th year grade of higher course; age: 12-21) executed in February, 1935. Every child pronounces the 100 Japanese basic syllables written on as many cards, which are exposed successively in a haphazard order. The following table shows the 100 Japanese basic syllables (kana) with their phonetic signs (in approximation) in Japanese arrangement. A number of young psychologists who are not familiar yet with voices of deaf children record the pronunciation. Five of them estimate every articulation of children in 5 grades : 4, 3, 2, 1, 0, (4 indicating very good andO cowmrpolnegtely or false) (A-group). The others record the pronunciations as they hear either with Japanese letter (kana) (B-group, 3 or 4 recorders) or with phonetic sign (C-group, 1 or 2 recorders).When we sum up the points of 5 recorders of A-group and average them, we can get the total point of 100 basic syllables of an individual and also the total point of 166 children of a basic syllable. Table 1 in the text (See p. 558) shows the distribution of total points of individuals (100 being the full mark). They have little correlation with the year grade nor with intelligence. Fig. 1 (See p. 559 in the Text) shows the distribution of the average point of each basic syllable in Japanese arrangement (20 being the full mark) (cf. the table of 100 Japanese basic syllables).Fig. 2 (See p. 560 in the Text) shows the percentage of cases, in which we find the same letter in the records of B-group as one on the original card (that is, the pronunciation was so good that the recorder could hear it as the original syllable,) with respect to each basic syllable. Both Fig. 1 and 2 show the same tendencies, above all, the W-like form almost in every line.We may analyse?a Japanese basic syllable (excepting 5 vowels) in two parts, the part of vowel and that of consonant. Table 3 in the text shows (See p. 561) the relation between the vowel part of original syllables (on the cards) and the vowel part of the records (of B-group). When a deaf child pronounces i, we hear u in many cases (when ki, we hear ku), when he pronounces e we hear it as a (when ke as ka), etc. We may find here an explanation of W-form in Fig. 1 and 2.We show the relation between the consonant part of the original syllables and that of the records in Tables 5-19 (See p. 563-64 in the Text) and summarise it in Table 20 (See p. 566 in the Text). The column on the left side in Table 20 shows the consonants of original syllables, letters in the upper line are consonants of syllables in the records. We may establish the grade of relationship (in a certain sense) between consonants in this table, for example, p-b,z-s-t,etc.