Search

In this work, six voiced/unvoiced speech classifiers based on the autocorrelation function (ACF), average magnitude difference function (AMDF), cepstrum, weighted ACF (WACF), zero crossing rate and energy of the signal (ZCR‐E), and neural networks (NNs) have been simulated and implemented in real time using the TMS320C6713 DSP starter kit. These speech classifiers have been integrated into a linear‐predictive‐coding‐based speech analysis‐synthesis system and their performance has been compared in terms of the percentage of the voiced/unvoiced classification accuracy, speech quality, and computation time. The results of the percentage of the voiced/unvoiced classification accuracy and speech quality show that the NN‐based speech classifier performs better than the ACF‐, AMDF‐, cepstrum‐, WACF‐ and ZCR‐E‐based speech classifiers for both clean and noisy environments. The computation time results show that the AMDF‐based speech classifier is computationally simple, and thus its computation time is less than that of other speech classifiers, while that of the NN‐based speech classifier is greater compared with other classifiers.

A method for spectral and spatial multichannel analysis/synthesis of interior aircraft sounds is presented. We propose two extensions of the classical sinusoids+noise model, adapted to multichannel stationary sounds. First, a spectral estimator is described, using average information across channels for spectral peak detection. Second, the residual modeling is extended to integrate two interchannel spatial cues (i.e., coherence and phase difference). This approach allows real-time synthesis and control of sounds spectral and spatial characteristics. It finds applications for multichannel aircraft sound reproduction, and more generally for musical and environmental sound synthesis. The ability of the model to reproduce multichannel aircraft sounds is assessed by a numerical simulation.

This letter proposes an HMM-based Text-to-Speech (TTS) system using waveform interpolation (WI)-based speech analysis and synthesis. The synthesized speech quality of the proposed system is significantly improved due to adopting an enhanced excitation modeling technique. The decomposition of characteristic waveform (CW) into slowly evolving waveform (SEW) and rapidly evolving waveform (REW) is efficient not only for excitation modeling but also for training process of HMMs. Objective and subjective test results verify the superiority of the proposed approach to conventional ones.

This paper introduces an analysis/synthesis scheme for the reproduction of sounds generated by sustained contact between rigid bodies. This scheme is rooted in a Source/Filter decomposition of the sound where the filter is described as a set of poles and the source is described as a set of impulses representing the energy transfer between the interacting objects. Compared to single impacts, sustained contact interactions like rolling and sliding make the estimation of the parameters of the Source/Filter model challenging because of two issues. First, the objects are almost continuously interacting. The second is that the source is generally unknown and has therefore to be modeled in a generic way. In an attempt to tackle those issues, the proposed analysis/synthesis scheme combines advanced analysis techniques for the estimation of the filter parameters and a flexible model of the source. It allows the modeling of a wide range of sounds. Examples are presented for objects of various shapes and sizes, rolling or sliding over plates of different materials. In order to demonstrate the versatility of the approach, the system is also considered for the modeling of sounds produced by percussive musical instruments.

This is the peer reviewed version of the following article: Julio C. Brégains; Francisco Ares "Analysis, synthesis, and diagnostics of antenna arrays through complex-valued neural networks", Microwave and Optical Technology Letters, 1512 - 1515 Volume: 48, Issue: 8, Aug. 2006, which has been published in final form at DOI: 10.1002/mop.21706. This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Self-Archiving [Abstract] It is shown in this paper that when artificial neural networks are extended to be complex valued, they can be incorporated as a very powerful and effective tool in the analysis, synthesis, and diagnostics of antenna arrays.

International audience; Vegetation analysis and reconstruction of remote sensing data are challenging problems. Using architectural tree models combined with model inputs estimated from aerial image analysis, this paper presents an analysis-synthesis approach for urban vegetation detection, modeling, and reconstruction. Tree species, height, and crown size information are extracted by aerial image analysis. These variables serve for model inversion to retrieve plant age, climatic growth conditions, and competition with neighbors. Functional-structural individual-based tree models are used to reconstruct and visualize virtual trees and their time evolutions realistically in a 3-D viewer rendering the models with geographical coordinates in the reconstructed scene. Our main contributions are: 1) a novel approach for generating plant models in 3-D reconstructed scenes based on the analysis of the geometric properties of the data, and 2) a modeling workflow for the reconstruction and growth simulation of vegetation in urban or natural environments

This paper investigates shape estimation of articulated 3D objects for object-based analysis-synthesis coding based on the source model of ‘moving articulated 3D objects’. For shape estimation three steps are applied: shape-initialization, object-articulation and shape-adaptation. Here, a new algorithm for object-articulation is introduced. Object-articulation subdivides a rigid model object represented by a mesh of triangles into flexibly connected model object-components. For object articulation, neighboring triangles which exhibit similar 3D motion during the image sequence are clustered into patches. These patches are considered to be the model object-components. For 3D motion estimation of a single triangle, a more reliable algorithm is proposed. The reliability is measured by the probability of convergency to correct parameters. For improving the reliability both, a more robust technique is applied and the triangle and its neighborhood are evaluated by the estimation algorithm. For clustering, a frame to frame clustering method which considers clustering results obtained in previous frames is presented. The developed algorithm for object-articulation is incorporated in OBASC. Typical videophone test sequences were applied. Compared to OBASC based on the source model of ‘moving rigid 3D objects’, the transmission rate decreases from 63.5 to 53 kbit/s at a fixed image quality measured by SNR. Furthermore, a realistic object articulation of a model object ‘body’ into object-components ‘head’ and ‘shoulders’ can be achieved without a priori knowledge about the scene content.

The topic of investigation was the extension of a generic object-based analysis-synthesis coder (OBASC) using the source model of ‘moving flexible 3D objects’ for the encoding of moving images at very low data rates to a knowledge-based coder in case scene-specific knowledge can be acquired. According to the coding concept, the OBASC describes and encodes each moving object of an image sequence by three parameter sets defining its motion, shape and surface color. The parameter sets of each object are obtained by image analysis. Using the coded parameter sets, an image can be synthesized by model-based image synthesis. The coder switches to a knowledge-based coder as soon as scene-specific knowledge can be acquired. In the example given in this paper, an algorithm tries to detect faces within moving objects using template matching and feature extraction techniques. If a face is detected, the face model Candide is adapted to the image sequence and integrated into the 3D model object describing the moving object. Due to this knowledge of the scene, the head and shoulders scene can be more efficiently modeled and encoded. Furthermore, the knowledge of the scene contents is used for controlling the coder. When compared to OBASC, the knowledge-based coder reduces the bit-rate required for encoding head and shoulder scenes by 17%.

The topic of investigation was object-based analysis-synthesis coding (OBASC) using the source model of ‘moving rigid 3D objects’ for the encoding of moving images at very low data rates. According to the coding concept, each moving object of an image is described and encoded by three parameter sets defining its motion, shape and surface color. The parameter sets of each object are obtained by image analysis. They are coded using an object-dependent parameter coding. Using the coded parameter sets, an image can be synthesized by model-based image synthesis. In comparison to block-based hybrid coding, OBASC requires the additional transmission of shape parameters. The transmission of shape information avoids the mosquito and block artifacts of a block-based coder. Furthermore, important areas such as facial areas can be reconstructed with a significant image quality improvement. OBASC based on the source models of ‘moving flexible 2D objects’ and of ‘moving rigid 3D objects’ gives almost identical image quality for the same data rate. Therefore the use of more advanced source models like flexible 3D objects or 3D face models is expected to further improve image quality.

A new method of identification, based on an input synchronized exponential swept-sine signal, is used to analyze and synthesize nonlinear audio systems like overdrive pedals for guitar. Two different pedals are studied; the first one exhibiting a strong influence of the input signal level on its input/output law and the second one exhibiting a weak influence of this input signal level. The Synchronized Swept Sine method leads to a Generalized Polynomial Hammerstein model equivalent to the pedals under test. The behaviors of both pedals are illustrated through model-based resynthesized signals. Moreover, it is also shown that this method leads to a criterion allowing the classification of the nonlinear systems under test, according to the influence of the input signal levels on their input/output law.

A source coding technique using multiple excitation pulses is expected to improve the quality of rule-based synthetic speech. In this paper, residual- and multipulse-excited analysis-synthesis methods are compared by evaluating the relationship between speech quality and bit rates. A method is proposed for reducing bit rates by thinning out prediction residual pulses (Thinned-Out Residual method). At excitation pulse utilization rates greater than 50 percent, this residual-excited analysis synthesis method produces higher-quality (segmental SNR) speech than the multipulse-excited analysis-synthesis method. When high-quality synthetic speech is required, the optimal residual utilization rate therefore is 100 percent. When a low bit rate is required, however, the pulse number should be reduced greatly. The multipulse-excited analysis-synthesis method then synthesizes speech with a higher segmental SNR.

A formal proof that bit allocation results hold for the entire class of paraunitary subband coders is presented. The problem of finding an optimal paraunitary subband coder, so as to maximize the coding gain of the system, is discussed. The bit allocation problem is analyzed for the case of the paraunitary tree-structured filter banks, such as those used for generating orthonormal wavelets. The even more general case of nonuniform filter banks is also considered. In all cases it is shown that under optimal bit allocation, the variances of the errors introduced by each of the quantizers have to be equal. Expressions for coding gains for these systems are derived. >

The author proposes an effective design structure for an analysis/synthesis filter bank with high computational efficiency. It can be shown that the resulting system filters are very efficient in terms of computational complexity and are relatively easy to design. Moreover, this design approach can be imposed on the system with relatively low reconstruction delays.

This paper proposes the pitch synchro nous mel inverse LSP analysis-synthesis technique as a new coding method for LPC voiced residual. the pitch synchronous amplitude spectrum of the voiced residual has characteristic zeros. By applying the inverse LSP analysis in mel domain to this spectrum, the spectrum can be encoded with a high efficiency with order of analysis as small as 12. Then mel LSP inverse filter is proposed which can synthesize directly the voiced residual in the time domain from the mel LSP coefficients. It is shown that the voiced residual can easily be synthesized. For the voiced residual obtained from the forementioned analysis-synthesis system, the order of the model formel inverse LSP analysis, quantization characteristics and the interpolation performance were investigated, with the mean spectrum distortion of mel pitch synchronous amplitude spectrum as the evaluation measure. the optimization of each parameter was discussed. A pitch synchronous mel inverse LSP analysis-synthesis system was constructed based on the obtained result. A subjective quality evaluation was made for the synthesized speech, and it was verified that a high-quality speech can be synthesized at 4.8 kbit/s. These results indicate that the proposed method is applicable with merits to the low bit-rate vocoder.

The performance of several algorithms for the quantization of the LSP (line spectrum polar) parameters is studied. An adaptive method that utilizes the ordering property of the LSP parameters is introduced. A combination of this adaptive algorithm with nonuniform-step-size quantization is shown to be very effective for encoding the LSP parameters. The performance of the different quantization schemes is studied on several sequences of speech samples. For the spectra distortion measure, appropriate performance comparisons between the different quantization schemes are rendered. >