Simple active noise-canceling systems (such as noise canceling headphones) use a feedback mechanism whereby the signal recorded by a microphone placed near the loudspeaker is phase-inverted and sent back through the loudspeaker via a feedback filter. The feedback filter is designed to achieve a high gain at low frequencies (for best noise reduction) while maintaining closed-loop stability under various conditions. To design the feedback filter, fairly ad-hoc techniques were traditionally used under the broad denomination of "loop-shaping" until the advent in the 1980s of optimal Hinfinalgorithms. The design technique outlined in this letter uses a cepstral domain approach, where gain and phase constraints take a convenient linear form, and linear programming to design the unique optimal feedback filter

Simple active noise-canceling systems (such as noise canceling headphones) use a feedback mechanism whereby the signal recorded by microphone placed near the loudspeaker is phase-inverted and sent back through the loudspeaker via a feedback filter. Such systems can be implemented using analog circuitry, are fairly inexpensive, and achieve reasonably good performance at low frequencies (below 500 Hz). The feedback filter is designed to achieve a high gain at low-frequencies (for best noise reduction) while maintaining closed-loop stability under various conditions. To design the feedback filter, fairly ad-hoc techniques were traditionally used under the broad denomination of "loop-shaping" until the advent in the 80's of optimal Hinfinalgorithms. The design technique outlined in this paper shares the intuitive approach of traditional loop-shaping methods, but removes most of the guess-work while producing optimal filters that guarantee closed-loop stability. The main idea is to represent the filter in the cepstral domain, where gain and phase constraints take a convenient linear form, and use linear programming to design the optimal feedback filter

We consider the problem of segmenting an audio signal into characteristic regions based on feature-set similarities. In the proposed approach, a feature-space representation of the signal is generated; sequences of these feature-space samples are then aggregated into clusters corresponding to distinct signal regions. The algorithm consists of using linear discriminant analysis (LDA) to condition the feature space and dynamic programming (DP) to identify data clusters. We consider the design of the dynamic program cost functions; we are able to derive effective cost functions without relying on significant prior information about the structure of the expected data clusters. We demonstrate the application of the LDA-DP segmentation algorithm to speech/music discrimination. Experimental results are given and discussed.

Additive synthesis is a powerful tool for the analysis/modification/synthesis of complex audio or speech signals. However, the cost of wavetable sinusoidal synthesis can become prohibitive for large numbers of sinusoids (more than a few hundred). In that case, techniques based on the inverse Fourier transform offer an attractive alternative, being 200-300% more efficient than wavetable synthesis depending on the number of sinusoids. This paper presents an improved technique based on the concatenation of short-term signals obtained by inverse Fourier transforms. In contrast to the standard overlap-add technique, the new algorithm requires synthesizing sinusoids in the frequency domain whose time-domain amplitudes vary linearly within the synthesis frame. The technique is shown to achieve higher quality than the standard overlap-add technique, at the cost of a small increase in computation.

The phase vocoder is a well established tool for time scaling and pitch shifting speech and audio signals via modification of their short-time Fourier transforms (STFTs). In contrast to time-domain time-scaling and pitch-shifting techniques, the phase vocoder is generally considered to yield high quality results, especially for large modification factors and/or polyphonic signals. However, the phase vocoder is also known for introducing a characteristic perceptual artifact, often described as "phasiness", "reverberation", or "loss of presence". This paper examines the problem of phasiness in the context of time-scale modification and provides new insights into its causes. Two extensions to the standard phase vocoder algorithm are introduced, and the resulting sound quality is shown to be significantly improved. Moreover, the modified phase vocoder is shown to provide a factor-of-two decrease in computational cost.

This paper deals with the application of short-time spectral attenuation techniques to the restoration of musical recordings degraded by background noise. Signal distortions induced by the restoration process are evaluated analytically, and their audibility is assessed on the basis of objective criteria. The results obtained highlight the influence of adjustable parameters (e.g., short-time frame duration or noise overestimation) on the quality of the restoration.>

Discusses models for sounds of percussive musical instruments such as vibraphones, pianos, and the like. In addition to classical source/filter models, a "multichannel excitation/filter model" is proposed in which a single excitation is used to generate several sounds, for example, five piano tones belonging to the same octave. Techniques for estimating the model parameters are presented along with their application to the sound of a real piano. The experiments demonstrate that it is possible to calculate a single excitation signal that, when fed into different filters, generates very accurate synthetic tones. Finally, a low-cost synthesis method that can be used to generate very natural sounding percussive tones is also proposed.>

A theoretical aspect of Prony's algorithm is presented, and the choice of analysis parameters and the achieved spectral resolution are briefly discussed. The analysis system itself is described, including frequency detection, component matching, and amplitude estimation. The problems connected with the analysis of complex signals are explored and some solutions proposed and developed. Three examples of musical sound analysis are then presented. It is concluded that the use of Prony's method to model musical sounds is efficient when applied to highly damped signals: its very good frequency resolution makes it possible to use small analysis windows and to follow efficiently the evolution of each sinusoidal component. The algorithm proposed for computing amplitudes avoids the problems due to biases in damping factor estimation. In the case of complex sounds, the analysis performed after presegmentation of the signal also yields good results. Future developments in this field are discussed.>