Speech Analysis Synthesis and Perception

Flanagan, James L.

doi:10.1007/978-3-662-01562-9

Cited by 1,083 publications

(458 citation statements)

References 0 publications

Supporting

Mentioning

404

Contrasting

Unclassified

Order By: Relevance

“…Figure 4 shows the realized group delay (solid line) and the ideal group delay (dashed line), where 0 ¼ 12 ms, f c ¼ 1;000 Hz and c e ¼ 1 are specified in Eq. (1). Note that the realized group delay is corrected to remove the delay component associated with all of the realized group delays in this study.…”

Section: Stimuli Designmentioning

confidence: 99%

“…In most analysis-synthesis systems, zero-or minimumphase reconstruction of speech is implemented [1] based on the assumption that, in human auditory perception, phase information is not very important [2]. Moriya and Honda [3] proposed a speech coder using phase equalization in which low-bit-rate compression and high-quality speech was obtained.…”

Section: Introductionmentioning

confidence: 99%

“…1 The stimuli are completely different. stimuli with a small group delay (À1 ms < 0 < 2 ms) were perceived to be similar under all combinations of f c and c e conditions.…”

mentioning

confidence: 99%

See 2 more Smart Citations

The effect of group delay spectrum on timbre.

Banno

Takeda

Itakura

2002

Acoust. Sci. & Tech.

View full text Add to dashboard Cite

This paper describes the results of subjective experiments investigating the characteristics of human phase perception. In these experiments, the stimuli have a flat amplitude spectrum and a group delay spectrum that has a single peak described by a Gaussian function where the mean (center frequency) and the standard deviation (bandwidth) are free parameters of interest. The first experiment was performed using stimuli with different peak values, where the center frequencies are fixed at 1,000 Hz and 4,000 Hz, and the bandwidths are also fixed. These bandwidths are computed by weighting the equivalent rectangular bandwidth (ERB) with certain parameters. It was found that when the peak values of the stimuli are between À1 ms and 2 ms, they are perceived to be zero phase, regardless of their center frequencies and bandwidths. Moreover, when the peak values of the stimuli are less than À8 ms or more than 10 ms and the bandwidths are less than 1 ERB, they are perceived to be similar. This result indicates that the nonzero-phase stimuli are similar when the bandwidth is less than 1 ERB. It was also found that the similarity of the stimuli reduces when the bandwidths of the nonzero-phase stimuli are more than 1 ERB.

show abstract

Section: Stimuli Designmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

The effect of group delay spectrum on timbre.

Banno

Takeda

Itakura

2002

Acoust. Sci. & Tech.

View full text Add to dashboard Cite

show abstract

“…These algorithms are generally based on a modeling of the vocal tract and do not work well for nonspeech audio signals. [3,4] The federal standards 1015 LPC (linear predictive coding) and 1016 CELP (coded excited linear prediction) fall into this category of audio compression. The -law transformation is a basic audio compression technique specified by the Comité Consultatif Internationale de Télégraphique et Téléphonique (CCITT) Recommendation G.711.…”

Section: Digital Audio Datamentioning

confidence: 99%

Digital Audio Compression

Mandal

2003

Multimedia Signals and Systems

View full text Add to dashboard Cite

Compared to most digital data types, with the exception of digital video, the data rates associated with uncompressed digital audio are substantial. Digital audio compression enables more efficient storage and transmission of audio data. The many forms of audio compression techniques offer a range of encoder and decoder complexity, compressed audio quality, and differing amounts of data compression. The -law transformation and ADPCM coder are simple approaches with low-complexity, low-compression, and medium audio quality algorithms. The MPEG/audio standard is a high-complexity, highcompression, and high audio quality algorithm. These techniques apply to general audio signals and are not specifically tuned for speech signals.

show abstract

“…But, a knowledge of the variation of Q with time t and of the corresponding mechanical state of the upper tract is sufficient to determine the principal properties of voiced speech. This can be achieved by means of a reduced complexity equation for Q(t) called the Fant equation [45][46][47], a nonlinear integro-differential equation deduced with the help of an aeroacoustic Green's function [48,49]. The equation is usually tractable, however, only when it can be assumed that the relevant acoustic frequencies are small enough for sound to propagate one-dimensionally in the vocal tract.…”

Section: Introductionmentioning

confidence: 99%