Download Musical Sound Effects in the SAS Model
Spectral models provide general representations of sound in which many audio effects can be performed in a very natural and musically expressive way. Based on additive synthesis, these models control many sinusoidal oscillators via a huge number of model parameters which are only remotely related to musical parameters as perceived by a listener. The Structured Additive Synthesis (SAS) sound model has the flexibility of additive synthesis while addressing this problem. It consists of a complete abstraction of sounds according to only four parameters: amplitude, frequency, color, and warping. Since there is a close correspondence between the SAS model parameters and perception, the control of the audio effects gets simplified. Many effects thus become accessible not only to engineers, but also to musicians and composers. But some effects are impossible to achieve in the SAS model. In fact structuring the sound representation imposes limitations not only on the sounds that can be represented, but also on the effects that can be performed on these sounds. We demonstrate these relations between models and effects for a variety of models from temporal to SAS, going through well-known spectral models.
Download Realtime Control of Audio Effects
Many sound processing effects need or can benefit from realtime control of one or more parameters. During interaction with the processor optimum settings can be achieved, or settings can be signal dependent. Time variance is a keyword here. For some effects the routing and mapping of the controller signals is of higher complexity than the audio signal routing itself. A concept for controlling a collection of effect blocks in real time with direct user interaction, and including controllers derived from the signal itself, is presented. This consists of a hierarchy of physical controllers, logical controllers, connection matrix, mapping functions, and effect parameters. And below that an effect parameter often controls several parameters in the signal processing algorithm. One controller can control more than one parameter and the mapping function from controller to the parameter can be set individually. Making this very flexible system easy to use presents some interesting challenges for the user interface design. Furthermore, as control signals can be derived from the audio signal itself, the signal processor has to calculate parameters from controller settings to a much higher extent than typically done, when a host processor takes care of this. The concept has been implemented in commercially available stand-alone effects processors.
Download Wavelet based Method for audio-video synchronization in broadcasting applications
The difference between standards used for films and for video generates problems when a conversion from one format to another is required : Since all the images are displayed, the change of frame rate induces a pitch change on the sound. To avoid this problem, the whole soundtrack has to be processed during the duplication. In this paper, we address the corresponding sound transformation problem, namely the dilation of the sound spectrum without changing its duration. For broadcasting applications, the ratio of transposition is within the range 24/25-25/24. The wide variety of sounds (music, speech, noise…) used in movies led us to first construct a database of representative sounds containing both transient, noisy and quasiperiodic sounds. This database has been used to compare the performances of different approaches. The reviewing of the most well known methods clearly shows significant disparities between them according to the class of the signal. This led us to reconsider the problem and to propose methods based on wavelet transforms.
Download Dynamic Models of Pseudo-Periodicity
Voiced musical sounds have non-zero energy in sidebands of the frequency partials. Our work is based on the assumption, often experimentally verified, that the energy distribution of the sidebands is shaped as powers of the inverse of the distance from the closest partial. The power spectrum of these pseudo-periodic processes is modeled by means of a superposition of modulated 1/f components, i.e., by a pseudo-periodic 1/f –like process. Due to the fundamental selfsimilar character of the wavelet transform, 1/f processes can be fruitfully analyzed and synthesized by means of wavelets, obtaining a set of very loosely correlated coefficients at each scale level that can be well approximated by white noise in the synthesis process.
Our computational scheme is based on an orthogonal P-band filter bank and a dyadic wavelet transform per channel. The P channels are tuned to the left and right sidebands of the harmonics so that sidebands are mutually independent. The structure computes the expansion coefficients of a new orthogonal and complete set of Harmonic Wavelets. The main point of our scheme is that we need only one parameter in order to model the stochastic fluctuation of sounds from a pure periodic behavior.
Download Filter Implementation On Synthup
This paper presents different implementations of digital filters on SYNTHUP, a PCI plug-in board based on FPGAs (Field Programmable Gate Arrays). The modular architecture of the board features a PCI interface, seven identical cells dedicated to computing, five FPGAs dedicated to data exchange, one cell for control and a total of 128M bytes of dual port SDRAM. The PCI interface offers bus mastering (scatter-gather DMA) for high speed data transfer between the board and the host memory. Applications are downloaded from the host via PCI into the logical resources of the FPGAs in few milliseconds. Each cell is made up of a FPGA that drives two independent 4M×16 SDRAM. It is connected to a 32 bit control bus and to an array of exchange FPGAs through a 64 bit bus. These FPGAs work as a crossbar switch. Moreover, they drive five 12 bit ports intended for communication with other boards. The paper concentrates on the implementation of FIR serial filters with distributed arithmetic which allows one FPGA to drive 48 channels in real time (48 kHz sampling rate, 16 bit data and coefficients). As the coefficients can be changed, each channel has its own filter (100 taps, adaptive filter). One can also build a 4800 taps filter for one channel.
Download A generalized 3-d resonator model for simulation of non rectangular shapes
A rectangular enclosure has such an even distribution of resonances that it can be accurately modeled using a feedback delay network, but a non rectangular shape such as a sphere has resonances that are distributed according to the extremal points of the spherical Bessel functions. This work proposes an extension of the already known feedback delay network structure to model a non rectangular shape such as a sphere. A speci c frequency distribution of resonances can be approximated, up to a certain frequency, by inserting an allpass lter of moderate order after the delay line within the comb lter structure. The feedback delay network used for rectangular boxes is therefore augmented with a set of allpass lters allowing parametric control over the enclosure size and the boundary properties. This work was motivated by informal listening tests which have shown that it is possible to identify a basic shape just from the distribution of its audible resonances.
Download Vibrato: detection, estimation, extraction, modification
This paper deals with vibrato detection, vibrato extraction on f 0 trajectory, and vibrato parameter estimation and modification. Vibrato detection and extraction are aimed at being a first step for note segmentation of singing voice signals. The aim is also to characterize sounds with the descriptor: "presence of vibrato" or "absence of vibrato". Changing vibrato parameters, that is to say its magnitude and its frequency, is also one of the possible musical applications. It is firstly required to detect the presence of vibrato. In order to do that, several approaches are possible: we can analyse directly the sound signal or its f 0 trajectory. For each approach, several techniques exist: some of them are described here: the "spectrum modelling" method, the "spectral envelopes distortion" method, the "AR prediction" method, the "analytic signal" method and the "minima - maxima detection" method. Their performance are compared. Secondly, the parameterization is completed: if there is vibrato, the parameters of the vibrato, that is to say its frequency and its magnitude, are given. Thirdly, the vibrato is extracted on f 0 trajectory to obtain a no-vibrato melodic evolution. This "flat" fundamental frequency is useful for segmentation of musical excerpts into notes, but can also be used for sound modification or processing.
Download Discrete-time Models for Non-linear Audio Systems
A variety of computational models have been proposed for digital simulation of nonlinear systems with memory [1, 2, 3, 4]. They are dealing with different aspects of the problem, like methods for identification, avoiding aliasing and fast convolution algorithms. In this paper we shortly sum up some of the common approaches and present a straightforward method for bandlimited discrete-time realization of analog nonlinear audio effects, like tube amps, exciters etc., using off-time digital cross correlation measurements. From these measurements we obtain a rather inefficient Wiener representation of the unknown nonlinearity. We then reduce the number of required coefficients significantly on the basis of multi-dimensional Laguerre transformation of the related Volterra kernels to allow real-time implementation on a digital signal processor .
Download Audio-rate control of FFT based processing using few parameters
Though the use of the Fast Fourier Transform (FFT) for signal processing in music applications has been widespread, applications in real-time systems for dynamic spectral transformation has been quite limited. The limitations have been largely due to amount of computation required for the operations. With faster machines, and with suitable implementations for frequency-domain processing, real-time dynamic control of high-quality spectral processing can be accomplished with great efficiency and simple approach. This paper will focus on dynamic real-time control of frequencydomain-based signal processing, and will describe the author's latest work (hi-resolution filtering and spatialization implementations) in this area. General background on the implementation and the development environment (Max Signal Processing, MSP) will also be provided.
Download Effect of early reflections in binaural systems with loudspeaker reproduction
Systems for 3D sound reproduction are often implemented with binaural technology where signals are played back over loudspeakers. This paper reports preliminary results from an investigation on how reflected sound in the listening room influences horisontal localisation in such systems. An experiment, consisting of listening tests, was done. Results from the experiment showed that reflections as late as 5ms and 10ms did influence localisation in such systems. The probability for reversals between front and back localisation increased, and the ability to localise to the back was degraded. Localisation was clustered towards the direction of the reflections.