This paper deals with problems of pitch-synchronous spectral analysis in the estimation of formants and bandwidths in connected speech. Pitch-synchronous analysis requires both the detection of the fundamental frequency of speech and the delineating of each period for analysis. This corresponds to a good determination of the starting point of a pitch period (i.e. the instant when the pitch pulse is applied to the speech production system), so that the pitch period could be thought of as a truncated (windowed) version of the impulse response of the vocal tract. If a nonharmonic spectral analysis is performed, by adding a suitable number of zero samples to the digitalized signal, the influence of a non-zero phase of the pitch pulse introduces, in the spectrum, undesired singularities which modify the values of the spectral maxima and bandwidths. These distortions are evaluated by representing the spectra in a 256 dimension vector space and introducing a suitable measure of the distance between elements of such a space. Digital spectra are computed by means of the F.F.T. and of the linear prediction in order to evaluate the influence of phase and windowing among different digital methods. The results of the tests for synthetic and natural speech are reported.
Click to purchase paper as a non-member or you can login as an AES member to see more options.
No AES members have commented on this paper yet.
To be notified of new comments on this paper you can
subscribe to this RSS feed.
Forum users should login to see additional options.
If you are not yet an AES member and have something important to say about this paper then we urge you to join the AES today and make your voice heard. You can join online today by clicking here.
