A new, low complexity method facilitates low burden embedding and recovery of tonal watermarks in speech. A watermark composed of a periodically extended sequence of sub-audible DTMF tones is added to speech asynchronously, without regard to momentary speech characteristics. It is detected through a combination of a bit manipulation enhancement and a data-directed correlation, ideal for simple hardware implementations. Three methods of bit manipulation enhancement were auditioned and the best selected for further investigation. It showed an average 26 dB processing gain vs. correlation alone, sufficient to detect the asynchronous sub-audible tones by a comfortable margin.

Interpolation of linear predictive coded (LPC) spectra is applied in a variety of speech processing tasks. We present a new method for interpolating between LPC spectra via pole-shifting. This approach solves the problem of real-to-complex and complex-to-real pole transitions by converting to a domain where each pole has a complex conjugate. Desired pole shifts are calculated in the new domain after applying a perceptually-based pole pairing algorithm. Intermediate spectra corresponding to these pole transitions are then optimally approximated using the original number of poles. The resulting interpolated spectral sequence is characterized by approximately linear changes in formant frequencies and bandwidths, and is free of the artifacts that may occur with other LPC spectral parameter interpolation methods.

Previously the authors proposed an adaptive spectral envelope modification method for the purpose of enhancing vocal tract resonance-disordered speech. This approach is to frequency warp abnormal vowel spectral envelope functions nonlinearly to match vowel spectra from a normal speaker; warping functions for nonvowel input spectra are estimated from spectrally similar vowel warping functions. A discussion is presented of attempts at optimizing the spectral warping accuracy and perceived output signal quality of the algorithm when applied to continuous speech modifications.>

The authors present the results of a study designed to investigate the effects of subtractive-type noise reduction algorithms on LPC-based spectral parameter estimation as related to the performance of speech processors operating with input SNRs of 15 dB and below. Subtractive noise preprocessing greatly improves the SNR, but system performance improvement is not commensurate. LPC spectral estimation is affected by the character of the residual noise which exhibits greater variance and spectral granularity than the original broadband noise. The study shows that removing less than the full amount of noise and whitening it improves spectral estimation and speech device performance. Techniques and performance results are presented.>

An algorithm is presented to determine the pitch of a signal in a very noisy environment adaptively and reliably. This algorithm is adaptive to input signal from frame to frame on both the correlative criterion and the energy criterion. The correlative criterion assumes that the input signal has a slowly time-varying property on adjacent frames. The energy criterion assumes that the input signal has some nonsignal frames to compute the statistics of noise energy. This algorithm has been useful in speech processing and speech coding. Implementation of this algorithm is simple and straightforward.>

A computationally efficient, although suboptimal, tree encoding method for the pitch and voicing parameters of an LPC (linear predictive coding) vocoder is presented. It is shown that when pitch and voicing signals are combined, it becomes difficult to take advantage of linear predictors and at the same time avoid any errors in voicing. The modified pitch tree coder presented here solves this problem by incorporating a branch leading to zero pitch value from each node of the tree. Only two bits per analysis frame are used to convey the combined pitch/voicing signal, with 1.585 bits being used to encode nonzero pitch values.>

Experimental results are reported in the area of speech modification for the purpose of correcting vocal tract resonance disorders. The linear predictive coding (LPC) coefficients and residual corresponding to an input speech signal are calculated, and a combination of nonlinear frequency warping and amplitude scaling is applied to produce modified LPC coefficients and then synthesize an output speech signal from these modified coefficients and the residual signal. The spectral envelope modification parameters are designed to adapt to the first two formant frequencies of the input speech. Cubic polynomial warping functions and amplitude scaling factors are found at vowel formant locations and are interpolated from these points over F1, F2 space. The results are compared with those obtained from frame-by-frame nonlinear spectral warping using a dynamic programming algorithm.>

A speech modification method presented by Griffin and Lim [1] involves modifying the Short-Time Fourier Transform Magnitude function of the original speech, and using it as a reference to iteratively approximate a modified waveform. This method has been successfully applied to time-scale modification and noise reduction of speech [1-3]. In this paper, we discuss our experiments in applying the algorithm to the independent modification of formant and pitch information.