Download Compression of sinusoidal modeling parameters
In this paper we present a technique for lossless compression of the sinusoidal modeling parameters. Compression is indeed useful for embedding spectral sounds in a synthesizer, broadcasting spectral sounds or simply storing many of them on a medium. This technique consists in compressing the frequency and amplitude parameters of each partial by adaptively sub-sampling and encoding their evolutions with time. It can be easily extended to handle spectral envelopes instead of partials, that is two-dimensional structures instead of one-dimensional ones. We have also designed a very compact file format for (compressed) spectral sounds based on this compression scheme.
Download Multichannel sound reproduction system for binaural signals - the ambisonic approach
Convincing sound reproduction via headphones requires filtering of virtual sound sources with head related transfer functions (HRTF). HRTFs describe signal differences at the two ear drums in level, time and frequency weighting [1,2]. Another psychoacoustic phenomenon of the human auditory system concerning hearing in natural sound fields is the improvement of source localization if small head movements are possible [3]. In this paper a computational efficient implementation to get dynamic binaural signals (in dependence on the head position) is proposed. This realtime modell is based on Ambisonic sound reproduction [4,5].
Download Acoustic test of a basic audio dereverberation process
Download Development of a surround encoding algorithm
The history of surround audio formats is examined, and a low cost solution to multi-channel encoding is explored and developed, taking DSP techniques into consideration . The algorithm developed will process four channels of audio and encode them so that a Dolby Surround decoder (or a Dolby ProLogic decoder) will successfully decode the signals when used as part of a home theatre / home entertainment system. The algorithm is a low cost alternative to the Dolby Encoding System.
Download Analysis, resynthesis and interpolation of the interior noise of a car
The aim of this work was to develop a realtime simulator for vehicle interior noise in varying load situations. Using this simulator, laboratory hearing tests which are common in the automotive industry may be executed more realistically than by using recorded sounds without any interaction possibility. A simplified model for the load situation of a car is presented. Using this model, the interpolation between the prerecorded signals is performed to obtain signals for other load situations. Furthermore, an analysis/resynthesis system for the interior noise of cars is introduced which only needs a sparse set of parameters.
Download Audio analysis by a model of the physiological auditory system
In this paper, an analysis of flute attacks processed by a model of the physiological auditory system is presented. In flute performance, the musician uses consonants (/d/, /g/, /k/, /t/ and /p/) in order to create particular effects as hard and soft attacks. These effects are very important in music interpretation. We found that the model discriminates the sounds very well and better than spectral analysis as worked out by standard methods. The model responses appeared very detailed and allowed attack classification by the application of very simple pattern recognition techniques.
Download Non-harmonic sinusoidal modeling synthesis using short-time high-resolution parameter analysis
This paper describes a method to perform analysis and resynthesis of non-harmonic and possibly simple polyphonic musical signals. Sinusoidal parameter extraction is performed using the highresolution matrix-pencil method. Instead of operating on the complete signal all at once, a short-time approach has been developed both to reduce the computational complexity and for the resynthesis to be capable of reproducing gradual frequency variations of partials. Partial tracking that utilizes a birth-death strategy is performed using the extracted frame based parameters. Resynthesis of signals with closely spaced beating partials are demonstrated.
Download High-level musical control paradigms for Digital Signal Processing
No matter how complex DSP algorithms are and how rich sonic processes they produce, the issue of their control immediately arises when they are used by musicians, independently on their knowledge of the underlying mathematics or their degree of familiarity with the design of digital instruments. This text will analyze the problem of the control of DSP modules from a compositional standpoint. An implementation of some paradigms in a Lisp-based environment (omChroma) will also be concisely discussed.