T. W. Parsons, "Separation of speech from interfering speech by means of harmonic selection," J. Acoust. Soc. Amer., vol. 60, pp. 911--918, 1976.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....problem arises when the (single) contaminating source is the speech of another talker. Not only is this a common situation in practice, but separation is likely to be maximally difficult since the target and contaminating signals will share obvious similarities. Early approaches to this problem [1] were monaural, estimating the fundamental frequency (or pitch ) of each talker ( f 1 0 and f 2 0 respectively) then selecting components of the frequency spectrum and assigning them to a talker according to their harmonic relation to the estimated pitch(es) This harmonic selection ....

Parsons, T.W. (1976) "Separation of speech from interfering speech by means of harmonic selection", Journal of the Acoustical Society of America, vol. 60, pp. 911--918.

....al. 1998a ] premised that the input audio signal contained just a single pitch sound or a single pitch sound with a noisy aperiodic sound. Although several methods for dealing with multiple pitch mixtures were proposed in the context of sound source segregation and automatic music transcription [ Parsons, 1976; Chafe and Ja#e, 1986; Katayose and Inokuchi, 1989; Brown and Cooke, 1994; Nakatani et al. 1995; Kashino and Murase, 1997; Kashino et al. 1998 ] they dealt with at most three musical instruments or voices and had great di#culty estimating the F0 in complex audio signals sampled from compact ....

Thomas W. Parsons. Separation of speech from interfering speech by means of harmonic selection. J. Acoust. Soc. Am., 60(4):911--918, 1976.

....asynchrony required in a similar range of tasks can vary from a few milliseconds for detection to several hundreds of milliseconds for tasks involving pitch and vowel identification. Models One of the earliest computational attempts at speech separation was the signal processing approach of Parsons (1976). Although Parsons was not motivated by auditory findings, his system served to define and partially solve some of the issues which have since become central for computational auditory scene analysis (CASA) systems operating on voiced speech, namely the resolution of overlapped harmonics, ....

....(1988) of two algorithms for the separation of voices based on a difference in fundamental frequency in a single channel. One approach operated by attenuating the pitch peak corresponding to the interfering voice through filtering the cepstrum of the mixed signal. The other was similar to Parsons (1976) harmonic selection scheme. By resynthesizing the target voice, possible speech enhancement benefits of these approaches could be evaluated. Stubbs and Summerfield used synthetic vowel pairs in one task and CV words masked by synthetic vowels in another to show that the enhanced 1999 Oct 25 8 ....

[Article contains additional citation context not shown here]

Parsons, T.W. (1976), Separation of speech from interfering speech by means of harmonic selection, Journal of the Acoustical Society of America, 60(4), 911-918.

....focused on feature extraction. In addition to the nascent CASA field, the specialized fields of speech recognition and music transcription and interpretation have bearing on the problem of acoustic stream segregation. Except for simultaneous multiple speaker speech segregation [Weintraub 1986, Parsons 1976] , most speech recognition work abstracted the problem of auditory scene analysis away by assuming that the acoustic signal either contained only speech or had a high signal tonoise ratio of speech to non speech sounds. Recently, work in speech recognition in noisy environments [Brown and Cooke, ....

Parsons, T. W., "Separation of speech from interfering speech by means of harmonic selection," Journal of the Acoustical Society of America, vol. 60, pp. 911--918, 1976.

....focused on feature extraction. In addition to the nascent CASA field, the specialized fields of speech recognition and music transcription and interpretation have bearing on the problem of acoustic stream segregation. Except for simultaneous multiple speaker speech segregation [Weintraub 1986, Parsons 1976] , most speech recognition work abstracted the problem of auditory scene analysis away by assuming that the acoustic signal either contained only speech or had a high signal tonoise ratio of speech to non speech sounds. Recently, work in speech recognition in noisy environments [Brown and Cooke, ....

Parsons, T. W., "Separation of speech from interfering speech by means of harmonic selection," Journal of the Acoustical Society of America, vol. 60, pp. 911--918, 1976.

.... hereafter referred to as the separation of periodic mixtures (SPM) problem, is therefore particularly relevant in reference to its application in the area of concurrent vowel or other periodic sound separation and in the process of separating a speech signal from an interfering speech signal [4] [5] as well as in reference to the interference rejection applications [3] 2] The SPM problem can be approached as a harmonic reassignment problem in the frequency domain [5] The FFT of the composite signal is dissected and the harmonics in the spectrum of the composite signal are tabulated and ....

.... or other periodic sound separation and in the process of separating a speech signal from an interfering speech signal [4] 5] as well as in reference to the interference rejection applications [3] 2] The SPM problem can be approached as a harmonic reassignment problem in the frequency domain [5]. The FFT of the composite signal is dissected and the harmonics in the spectrum of the composite signal are tabulated and tagged as belonging to either of the components. Time domain waveforms for either component are then obtained by using the inverse FFT [5] On the other hand, the SPM problem ....

[Article contains additional citation context not shown here]

T. W. Parsons, "Separation of speech from interfering speech by means of harmonic selection," Journal of Accoustical Society of America, vol. 60, pp. 911 -- 918, Oct. 1976.

....by the need for a front end processor for robust automatic speech recognition in noisy environments. Early work includes the system of Weintraub [57] which attempted to separate the voices of two speakers by tracking their fundamental frequencies (see also the nonauditory work of Parsons [40]) More recently, a number of multistage computational models have been proposed by Cooke [12] Mellinger [35] Brown and Cooke [7] and Ellis [16] Generally, these systems process the acoustic input with a model of peripheral auditory function, and then extract features such as onsets, offsets, ....

T. W. Parsons, "Separation of speech from interfering speech by means of harmonic selection," J. Acoust. Soc. Am., vol. 60, no. 4, pp. 911--918, 1976.

....the identity of a second vowel. When they differ, the stimulus sounds like two talkers pronouncing different vowels with different pitches. A number of models or algorithms have been proposed to account for, or reproduce, our ability to separate speech using a difference in fundamental frequency (Parsons 1976; Frazier, Samsam, Braida and Oppenheim 1976; Nagabuchi, Kobayashi and Yamamoto 1979; Scheffers 1983; Kitamori, Harada and Kawarada 1984; Weintraub 1985, 1986; Palmer 1988, 1990; Stubbs and Summerfield 1988, 1990; Assmann and Summerfield 1990; Duda and Lyon 1990; Meddis and Hewitt 1990) Most of ....

Parsons, T. W. (1976). "Separation of speech from interfering speech by means of harmonic selection," J. Acoust. Soc. Am. 60, pp. 911-918.

....a single source. Classic double vowel experiments are blind to to this aspect because they force listeners to report two vowels for every trial. We use a modified task in which a subject may report one or two vowels. F0 guided segregation has been explained by models inspired from Parsons work [24], and models in the vein of Weintraub s system [27] They differ in the degree of selectivity required of peripheral analysis: sufficient to resolve individual components for the former, sufficient to isolate spectral zones reflecting each vowel for the latter. Assmann and Summerfield [1] argued ....

Parsons, T. W. (1976). "Separation of speech from interfering speech by means of harmonic selection." J. Acoust. Soc. Am. 60, 911-918.

....the set of channels and fundamental frequency are refined for each source in alternation, until there is no further change in the state of the network. Other approaches to source separation which progressively refine the F0 estimate for each source have been described by a number of workers (Parsons, 1976; Scheffers, 1983; Lea, 1992; de Cheveign , 1993) although not in the context of a neural network architecture. These approaches share the same rationale as the model described here; namely, that it is easier to estimate the characteristics of one source in an acoustic mixture when the other ....

Parsons, T.W. (1976) Separation of speech from interfering speech by means of harmonic selection.

....the generation of new processing streams, the formalization of this KS would represent an attempted answer to the general problem of modeling acoustic source interaction. In other acoustic domains such as speech recognition, this problem manifests itself in simultaneous multiple talker scenarios [32] and in the coarticulation effect [24] 6 of connected speech. It is true that the standard numeric level approach to reprocessing via selective processing streams might work here and thus eliminate the need for model synthesis. One could conceivably filter out sources to find evidence for ....

T. W. Parsons, "Separation of Speech from Interfering Speech by Means of Harmonic Selection," Journal of the Acoustic Society of America, vol 60, no. 4, pp 911--918, Oct. 1976.

....(Rabiner, 1989; Bourlard Morgan, 1994) Obviously, auditory segregation is a necessary ability for any speech recognizer to work in a realistic environment. Thus, a successful method of auditory segregation would be a breakthrough in making speech recognition technology reach the real world. Parsons (1976) developed a computer program to separate two speakers on the basis of different fundamental frequencies. It uses Fourier analysis as the front end to extract frequency partials. Based on the extracted frequency partials, the algorithm computes the fundamental frequency of the first speaker that ....

....analysis, which is largely an unsolved engineering problem. Automatic auditory segregation is a critical part of auditory signal processing, real time speech recognition and music transcription in 20 20 natural environments. Compared to existing computer algorithms for auditory input separation (Parsons, 1976; Weintraub, 1986; Mellinger, 1992; Brown Cooke, 1994) the oscillatory correlation approach offers many unique advantages. Due to oscillatory dynamics, no single stream can dominate and suppress the perception of the rest of the auditory scene for a long time. The processing is inherently ....

Parsons, T.W. (1976). Separation of speech from interfering speech by means of harmonic selection.

....In other words, it could improve the robustness of J RASTA to compensate the spectrum before J RASTA in linear or logarithmic domains. Separation of speech from noisy signals based on typical structures of speech such as harmonic structure have been investigated, especially in speech enhancement [6, 7, 8, 9]. It is another way to compensate the corrupted speech. The possibility of improving ASR systems has been suggested, but has not been evaluated with ASR very much. We apply a harmonic sieving technique (HS) and a spectral subtraction (SS) technique as preprocessing of JRASTA feature extraction. ....

....features of the speech signal. However, in most ASR systems, this is smoothed out in order to reduce the variability of the speech spectrum. On the other hand, the harmonic structure has been investigated and used in speech enhancement, speech separation, and computational auditory scene analysis [6, 7, 8, 9]. These techniques can be effective for robust speech recognition, but they are not often evaluated in ASR system with real noise. We assume that the speech signal in the voiced region exists only in the harmonic structure. With this assumption, we can improve the signal to noise ratio by sieving ....

Thomas W.Parsons. Separation of speech from interfering speech by means of harmonic selection. The Journal of the Acoustical Society of America, 60(4):911--918, - 1976.

....by the need for a front end processor for robust automatic speech recognition in noisy environments. Early work includes the system of Weintraub [59] which attempted to separate the voices of two speakers by tracking their fundamental frequencies (see also the non auditory work of Parsons [43]) More recently, a number of multi stage computational models have been proposed by Cooke [12] Mellinger [36] Brown and Cooke [7] and Ellis [16] Generally, these systems process the acoustic input with a model of peripheral auditory function, and then extract features such as onsets, offsets, ....

T. W. Parsons, "Separation of speech from interfering speech by means of harmonic selection", J. Acoust. Soc. Am., vol. 60(4), pp. 911-918, 1976.

....MODEL AS AN ALTERNATIVE TO PERIODICITY METHODS A brief review of previous work in double vowel separation Vowel separation seems a tantalizingly tractable problem, since the presumed different fundamental frequencies should form a powerful basis for segregation. Such a technique was described by Parsons (1976), which sought to separatetwo voices by identifying both pitches andusing this to segregate the harmonics belonging to each talker in a narrow band short time Fourier transform. de Cheveign (1993) describes this and many subsequent models based both on that kind of frequency domain ....

Parsons, T. W. (1976). "Separation of speech from interfering speech by means of harmonic selection," JASA 60(4), 911-918.

No context found.

T. W. Parsons, "Separation of speech from interfering speech by means of harmonic selection," J. Acoust. Soc. Amer., vol. 60, pp. 911--918, 1976.

No context found.

T.W. Parsons. Separation of speech from interfering speech by means of harmonic selection. J. Acoust. Soc. Am., 60(4):911--8, October 1976.

No context found.

T. W. Parsons, "Separation of speech from interfering speech by means of harmonic selection," J. Acoust. Soc. Amer., vol. 60, pp. 911--918, 1976.

No context found.

T.W. Parsons. Separation of speech from interfering speech by means of harmonic selection. J. Acoust. Soc. Am., 60(4):911--8, October 1976.

No context found.

T.W. Parsons. Separation of speech from interfering speech by means of harmonic selection. J. Acoustic Society of America, 60(4), 1976.

No context found.

Parsons, T. W. (1976). "Separation of speech from interfering speech by means of harmonic selection," J. Acoust. Soc. Am. 60, 911-918.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC