Wyniki wyszukiwania - BazTech

1

Methods of deformed voice signal evaluation after larynx surgery

Wszołek W., Kłaczyński M.

Mechanics / AGH University of Science and Technology

|

2009

|

Vol. 28, no. 1

31-37

EN

In the work has been shown from studies concerning the application of modified acoustic signal processing methods to the task of evaluation and classification of larynx surgery effects. The goal of the standard speech recognition studies is to reveal the semantic aspects of the pronounced text. In the tasks of medical diagnosis employing the speech signal analysis the semantic aspects are insignificant. The required signal characteristics should be as sensitive as possible to small deformations of the layers directly related to the voice functioning and the structure of vocal tract. The goal of the work is presentation of voice quality after various surgical treatments, performed in the ENT area. The research subject is the speech articulation process itself and all its pathological deformations, which determines both the used signal analysis tools as well as the techniques of the selected objects recognition, which are the forms of the particular ill person speech deformation forms in comparison to the speech of the whole sound people population. The evaluation has been carried out both for voice quality after larynx surgery as well as voice quality after surgical treatment of resonance cavities (nose, paranasal sinussis). The study was oriented towards the construction of systems based on the analysis of objectively registered acoustic signals of deformed speech.

PL

W pracy przedstawiono badania dotyczące metod przetwarzania sygnału akustycznego do oceny i klasyfikacji mowy po zabiegach w obrębie kanału głosowego. W zagadnieniach rozpoznawania mowy, problem dotyczy ujawniania semantycznych aspektów wypowiedzi. Natomiast w zagadnieniach diagnostyki medycznej przy wykorzystaniu sygnału mowy, cechy semantyczne są nieistotne. Poszukiwane cechy sygnału mowy winny być wrażliwe na małe deformacje, które mogą wystąpić w poszczególnych warstwach kanału głosowego. Celem pracy jest ocena jakości głosu po różnorodnych zabiegach chirurgicznych wykonanych w obszarze kanału głosowego. Tematem badań jest zarówno sam proces artykulacji mowy, jak i jego patologiczne deformacje. Diagnostykę narządu głosu można określić jako jednoznaczne rozpoznanie cech aktualnego stanu źródła głosu na podstawie zespołu istotnych cech akustycznych, zwartych w sygnale akustycznym. Ocena jakości głosu została przeprowadzona dla osób po chirurgicznym leczeniu krtani, nosa oraz zatok przynosowych. Badania zostały ukierunkowane na stworzenie systemu analizy umożliwiającego obiektywne rozpoznawanie deformacji sygnału mowy.

2

Study of effects of surgical treatment in the larynx area on the speech signal

Wszołek W., Kłaczyński M., Modrzejewski M.

Archives of Acoustics

|

2008

|

Vol. 33, No. 4

497-508

EN

The speech signal emitted by humans may be a source of useful diagnostic and prognostic information. The signal may become, indirectly by some selected parameters, an additional source of information concerning the condition of the patient's vocal tract anatomy, as well as physiology and pathology (deformation) of his/her other internal organs. The paper presents the next, consecutive stage of the authors' research, concerning the search for additional parameters, which could be used for objective detection and registration of pathological changes in the larynx and vocal tract area.

3

A Vibroacoustic Model of Selected Human Larynx Diseases

Engel Z. W., Kłaczyński M., Wszołek W.

International Journal of Occupational Safety and Ergonomics

|

2007

|

Vol. 13, No. 4

367--379

EN

With the present development of digital registration and methods for processing speech it is possible to make effective objective acoustic diagnostics for medical purposes. These methods are useful as all pathologies and diseases of the human vocal tract influence the quality of a patient’s speech signal. Diagnostics of the voice organ can be defined as an unambiguous recognition of the current condition of a specific voice source. Such recognition is based on an evaluation of essential acoustic parameters of the speech signal. This requires creating a vibroacoustic model of selected deformations of Polish speech in relation to specific human larynx diseases. An analysis of speech and parameter mapping in 29-dimensional space is reviewed in this study. Speech parameters were extracted in time, frequency and cepstral (quefrency) domains resulting in diagrams that qualified symptoms and conditions of selected human larynx diseases. The paper presents graphically selected human larynx diseases.

4

Automatic understanding of acoustic speech signal pathology

Wszołek W.

Archives of Acoustics

|

2007

|

Vol. 32, No. 1

147-163

EN

In this work, parts of the research concerning a new concept of applying computer technique in pathological speech analysis have been presented. This new concept assumes that during the pathological speech analysis we are not aiming neither at the establishing of such or other signal parameters nor at the trying to classify them, but we tend to understand automatically the causes of deformation, which can be observed in the considered signal. Therefore the concept presented postulates the replacing the well known process of the pathological speech acoustic signal recognition by a more advanced method of analysis, which means a confrontation of the features, which are revealed in the signal during its transformation with features that could be expected basing on the knowledge gathered in the system concerning pathological factors deforming the true form of the signal. In the meaning of the term "automated understanding", this denotes a signal analysis of a deformed speech, which is oriented towards revealing the sources of the observed signal distortions, and not towards bare analysis of their patterns and diagnostic deduction based on their typology. In the work the basic elements of the proposed method are presented. Examples showing its essence were derived basing on the selected larynx pathology analysis.

5

Application of new acoustic parameters in ANN-aided pathological speech diagnosis

Szaleniec J., Modrzejewski M., Szaleniec M., Wszołek W.

Archives of Acoustics

|

2007

|

Vol. 32, No. 1

177-186

EN

Most diseases of the vocal tract cause changes in the voice quality. Acoustic analysis of the speech signal is a widely used, noninvasive, objective and low-cost method of laryngeal pathology recognition and classification. There have been numerous attempts [1-3] to develop an automatic system which could aid the laryngological diagnosis. The goal of the presented research is to verify, whether the new approach to the acoustic analysis and parameters introduced in the Voice Analysis and Screening System (VASS 3.0 [4]) such as turbulence noise index (TNI) and normalized first harmonic energy (NFHE), can improve the effectiveness of automated diagnosis. The automated diagnosis was performed using Artificial Neural Networks (ANN). Multilayer perceptron and radial basis function neural networks of various architectures were trained to classify between pathologic and non-pathologic voices, while the parameters computed with VASS were used as input data. Preliminary results show that the Voice Analysis and Screening System coupled with ANN can be a highly effective tool for ANN-aided pathological speech diagnosis.

6

Selected methods of pathological speech signal analysis

Wszołek W.

Archives of Acoustics

|

2006

|

Vol. 31, No. 4

413-430

EN

Selected results of examinations, having been performed by the author for several years, concerning the evaluation of selected methods of transforming signal in analysis tasks and pathological speech evaluation usability were presented in the article. In many issues of medical diagnosis, as well as in the planning of some illnesses therapy and rehabilitation, it is necessary to evaluate the signal of deformed speech. This evaluation can concern evaluation of the signal deformation degree and in this case the task is to present in a quantitative dimension (preferably scalar) the measure of deviation between the measured pathological speech signal and the abstract signal, which can be acknowledged as a pattern of the correct speech. Possessing such a scalar measurement, of the considered signal deformation degree, enables to monitor the illness process development or remission, which has a key importance in the current monitoring of the therapy effects or/and many illnesses rehabilitation. Pathological speech evaluation can also go into the direction of a classification and determination of a type of the analysed signal, which can, in some cases, have a direct connection with anatomical and functional causes as well as conditions of the considered illness. The classification of pathological speech signals can facilitate the diagnosis by pointing the most probable causes of the speech signal pathological deformation. Such a classification can also help in the optimal therapy selection and in determining rehabilitation recommendations. The methods of transformation, analysis, classification and speech signal recognition have been known apparently for many years, as with no difficulty, many writings discussing these concepts and presenting results concerning both basic examinations results and many application works, can be found. The problem, presented in this article, is strongly different from the majority of works, which were published by other authors, as most of well known works, concerning the speech signal analysis (etc.), are directed to understand the content of the statement (automatic speech understanding, a speech-writing conversion, steering of the devices by speech signal), or relatively by determining the identity of the speaker (automatic identification or authentification of the speaker). Meanwhile, in the pathological speech analysis the semantic statement content is not essential; it is also not important who the speaker is. The research subject is the speech articulation process itself and all its pathological deformations, which determines both the used signal analysis tools as well as the techniques of the selected objects recognition, which are the forms of the particular ill person speech deformation forms in comparison to the speech of the whole sound people population [5].

7

Recognition of Pathological Utterances with the Usage of the Kohonen Neural Networks

Kapusta M., Gajer M., Shomali A.

Bulletin of the Polish Academy of Sciences. Technical Sciences

|

2000

|

Vol. 48, nr 1

117-122

EN

The nature of speech signal is very complicated, that causes that its visualization and further analysis, without some initial pre-processing, is very complicated and doesn't always bring the desired effects. Speech signal in most cases is represented by videograms. The analysis of these forms of signal visualisation is not easy because of difficulties in their interpretation. In this article the usage of Kohonen neural networks for visualising speech signals uttered by children with a cleft palate was proposed. Speech signal is converted to its spectrum matrices representation, which in turn constitutes the input for Kohonen neural networks. Further a method for generating a simplified form of speech signal ( a poly=line figure) based on the network's output was presented. In addition a method for pathological speech signal recognition was presented. Test results based on utterances obtained from children with a cleft palate were presented.

8

Rozpoznawanie mowy patologicznej na podstawie obrazów głosek szumowych

Kapusta M., Gajer M., Shomali A.

Kwartalnik Elektroniki i Telekomunikacji

|

1999

|

Vol. 45, nr 3-4

405-417

PL

Sygnał mowy posiada bardzo skomplikowaną naturę, która sprawia że jego zobrazowanie oraz dalsza analiza bez operacji wstępnego przetworzenia są trudne i nie zawsze przynoszą pożądane efekty. W wielu pracach sygnał mowy przedstawiany jest w postaci wideogramów, będących wykresami czasowo - częstotliwościowymi, jednakże analiza tych obrazów nie jest łatwa ze względu na ich trudną interpretacje. W pracy niniejszej zaproponowano wykorzystanie sieci neuronowej Kohonena do generacji obrazów sygnałów mowy patologicznej, występującej u dzieci z rozszczepem podniebienia. Opisano sposób przekształcenia sygnału mowy do postaci macierzy widm chwilowych, stanowiącej zbiór danych wejściowych dla układu sieci neuronowej Kohonena. Następnie omówiono metodę generacji obrazu przez sieć neuronową oraz zaprezentowano metodę identyfikacji mowy patologicznej na podstawie otrzymanych obrazów, opierającą się na pomiarze całkowitej długości linii łaczącej zwycięskie neurony. Otrzymane dla poszczególnych głosek rezultaty pomiarów długości linii zobrazowano w postaci wykresów.

EN

The nature of speech signal is very complicated, that causes that its visualisation and further analysis, without some initial pre - processing, is very complicated and doesn't always bring the desired effects. Speech signal in most cases is represented by videograms. The analysis of these forms of signal visualisation is not easy because of difficulties in their interpretation. In this article the usage of Kohonen neural network for visualising speech signals uttered by children with a cleft palate, was proposed. Speech signal is converted to its spectrum matrices representation, which constitutes the input for Kohonen neural network. Further a method for generating a simplified form of speech signal (a poly - line figure) based on the network's output, was discussed. In addition, a method for pathological speech signal recognition was proposed. Test results based on utterances obtained from children with a cleft palate were also presented.