Thesis (M. Eng.)--Carleton University, 1988. / Also available in electronic format on the Internet.

In der vorliegenden Dissertation wurden 24 normalhörende Probanden bezüglich ihres Musikverständnisses vor und nach spektraler Reduktion verschiedener Musiksignale untersucht. Die spektrale Reduktion wurde mittels Rauschbandvokoder vorgenommen und ergab drei unterschiedliche Hörkonditionen: 1. nicht reduzierte Originalvariante 2. reduzierte Variante mit zwölf Rauschbändern 3. reduzierte Variante mit sechs Rauschbändern Als Untersuchungsplattform diente der Mu.S.I.C-Test, der verschiedene Teilaspekte des Musikhörens und Musikverstehens mit folgendem Ergebnis untersucht: 1. Die Rhythmuswahrnehmung war durch spektrale Reduktion nicht beeinträchtigt. 2. Die Tonhöhenunterscheidungsfähigkeit nahm mit zunehmender spektraler Reduktion stark ab. 3. Der Melodietest ist eine andere Form eines Tonhöhentests und kam entsprechend zu einem ähnlichem Ergebnis; auch hier nahm die Erkennungsleistung mit spektraler Reduktion stark ab. 4. Die emotionale Beurteilung verschiedener Musikstücke war in den drei Hörkonditionen im Mittel gleich, ein gegenseitiger Bezug bei der Bewertung der Einzelstücke blieb nach spektraler Reduktion grob erhalten. Als dominierender Faktor wurde das Tempo der Musikstücke ausgemacht. 5. Den Dissonanztest beantworteten die Probanden in der Originalvariante entsprechend der pythagoräischen Konsonanztheorie. In den spektral reduzierten Konditionen konnte nur die Tonhöhe als Einflussfaktor bestätigt werden. Eine Korrelation zwischen Konsonanzempfinden und der An- oder Abwesenheit bestimmter Instrumente konnte nicht belegt werden. 6. Nach spektraler Reduktion war die korrekte Bestimmung der Instrumentenanzahl erschwert. 7. Beim Akkordvergleich erkannten die Probanden teilweise auch nach spektraler Reduktion subtile Unterschiede. Die Unterscheidung in "gleich"' oder "ungleich"' erscheint als zu leicht und konnte nicht zur weiteren Modellbildung beitragen. 8. Zunehmende spektrale Reduktion verminderte die Fähigkeit der Instrumentenerkennung. Häufig verwechselt wurden Intrumente, die in der Art der Tonerzeugung ähnlich sind. / In this work, 24 normal hearing subjects were tested regarding their appreciation of music before and after spectral reduction of different musical sounds. The spectral reduction was achieved by employing a noise band vocoder and resulted in three different listening conditions: 1. spectrally not reduced original condition 2. spectrally reduced condition with 12 noise bands 3. spectrally reduced condition with 6 noise bands The Mu.S.I.C test battery served as the primary research tool. It consists of a number of individual tests, which explore different aspects of hearing and understanding of music. Main outcomes of the investigation are: 1. Identification of rhythm wasn't affected by spectral reduction. 2. Ability to differentiate pitch drops significantly with increasing spectral reduction. 3. The melody test also investigates pitch perception and the results are comparable: the rate of correct melody identification drops significantly with increasing spectral reduction. 4. The emotional appreciation of different musical stimuli was on average equal in all conditions, however ranking was different. In both spectrally reduced cases, emotional ranking was roughly the same. Here, the dominating factor of appreciation was the tempo. 5. In the original condition, subjects answered the dissonance test according to the Pythagorean consonance theory. In the spectrally reduced conditions, only pitch could be identified as influencing factor. A correlation between the presence or absence of instruments and the perception of consonance could not be demonstrated. 6. The correct identification of the number of instruments was handicaped after spectral reduction. 7. After spectal reduction, subjects were in parts able to discriminate subtle differences when comparing chords. The task of judging "equal" and "different" appears too easy and wasn't suitable to contribute to further modeling. 8. The ability to identify instruments was reduced on increasing spectral reduction. Instruments with similar modes of sound generation were more likely to be confused.

The area of speech recognition by machine is one of the most popular and complicated subjects in the current multimedia field. Linear predictive coding (LPC) is a useful technique for voice coding in speech analysis and synthesis. The first objective of this research was to establish a prototype of the residual-excited linear predictive (RELP) vocoder system in a real-time environment. Although its transmission rate is higher, the quality of synthesized speech of the RELP vocoder is superior to that of other vocoders. As well, it is rather simple and robust to implement. The RELP vocoder uses residual signals as excitation rather than periodic pulse or white noise. The RELP vocoder was implemented with Texas Instruments TMS320C6711 DSP starter kit (DSK) using C. Identifying vowel sounds is an important element in recognizing speech contents. The second objective of research was to explore a method of characterizing vowels by means of parameters extracted by the RELP vocoder, which was not known to have been used in speech recognition, previously. Five English vowels were chosen for the experimental sample. Utterances of individual vowel sounds and of the vowel sounds in one-syllable-words were recorded and saved as WAVE files. A large sample of 20-ms vowel segments was obtained from these utterances. The presented method utilized 20 samples of a segment's frequency response, taken equally in logarithmic scale, as a LPC frequency response vector. The average of each vowel's vectors was calculated. The Euclidian distances between the average vectors of the five vowels and an unknown vector were compared to classify the unknown vector into a certain vowel group. The results indicate that, when a vowel is uttered alone, the distance to its average vector is smaller than to the other vowels' average vectors. By examining a given vowel frequency response against all known vowels' average vectors, individually, one can determine to which vowel group the given vowel belongs. When a vowel is uttered with consonants, however, variances and covariances increase. In some cases, distinct differences may not be recognized among the distances to a vowel's own average vector and the distances to the other vowels' average vectors. Overall, the results of vowel characterization did indicate an ability of the RELP vocoder to identify and classify single vowel sounds.

The area of speech recognition by machine is one of the most popular and complicated subjects in the current multimedia field. Linear predictive coding (LPC) is a useful technique for voice coding in speech analysis and synthesis. The first objective of this research was to establish a prototype of the residual-excited linear predictive (RELP) vocoder system in a real-time environment. Although its transmission rate is higher, the quality of synthesized speech of the RELP vocoder is superior to that of other vocoders. As well, it is rather simple and robust to implement. The RELP vocoder uses residual signals as excitation rather than periodic pulse or white noise. The RELP vocoder was implemented with Texas Instruments TMS320C6711 DSP starter kit (DSK) using C. Identifying vowel sounds is an important element in recognizing speech contents. The second objective of research was to explore a method of characterizing vowels by means of parameters extracted by the RELP vocoder, which was not known to have been used in speech recognition, previously. Five English vowels were chosen for the experimental sample. Utterances of individual vowel sounds and of the vowel sounds in one-syllable-words were recorded and saved as WAVE files. A large sample of 20-ms vowel segments was obtained from these utterances. The presented method utilized 20 samples of a segment's frequency response, taken equally in logarithmic scale, as a LPC frequency response vector. The average of each vowel's vectors was calculated. The Euclidian distances between the average vectors of the five vowels and an unknown vector were compared to classify the unknown vector into a certain vowel group. The results indicate that, when a vowel is uttered alone, the distance to its average vector is smaller than to the other vowels' average vectors. By examining a given vowel frequency response against all known vowels' average vectors, individually, one can determine to which vowel group the given vowel belongs. When a vowel is uttered with consonants, however, variances and covariances increase. In some cases, distinct differences may not be recognized among the distances to a vowel's own average vector and the distances to the other vowels' average vectors. Overall, the results of vowel characterization did indicate an ability of the RELP vocoder to identify and classify single vowel sounds.

No description available.

No description available.

No description available.

Thesis (M.S.)--Virginia Polytechnic Institute and State University, 1994. / Vita. Abstract. Includes bibliographical references (leaves 148-154). Also available via the Internet.

This thesis describes the subjective analysis of a DPCM system featur- n ing an adaptive quantizer. The system is simulated on a digital computer and operated under variations in the sampling frequency and the number of available quantizer levels. The subjective performance of the system is judged using the isopreference method which presents test results in the form of isopreference contours.drawn on a plane showing sampling frequency and number of quantizer levels as axes. From these curves the minimum required channel capacity for a given subjective preference level is shown to occur when sampling is at the Nyquist rate. The previous statement applies when the quantizer output levels are naturally coded or entropy coded. The isopreference contours indicate implementation tradeoffs between the number of quantizer levels and the sampling frequency. The isopreference contours also show that odd level quantizers outperform even level quantizers when entropy coding is used. Analytical measures of performance in the form of output signal-to-noise ratio (SNR) are obtained. Although correlation between curves of constant SNR and curves of constant subjective quality are evident, the SNR curves do not accurately reflect the results of subjective evaluation. A special experiment involving quantizer dc offset is described which indicates that SNR could not be used to compare speech samples containing large proportions of different types of noise. Throughout the work, the digital channel between encoder and decoder is assumed noiseless. / Applied Science, Faculty of / Electrical and Computer Engineering, Department of / Graduate

No description available.