Raveendra K. Rao

Audio Signal Feature Extraction and Classification Using Local Discriminant Bases

Audio feature extraction plays an important role in analyzing and characterizing audio content. Auditory scene analysis, content-based retrieval, indexing, and fingerprinting of audio are few of the applications that require efficient feature extraction. The key to extract strong features that characterize the complex nature of audio signals is to identify their discriminatory subspaces. In this paper, we propose an audio feature extraction and a multigroup classification scheme that focuses on identifying discriminatory time-frequency subspaces using the local discriminant bases (LDB) technique. Two dissimilarity measures were used in the process of selecting the LDB nodes and extracting features from them. The extracted features were then fed to a linear discriminant analysis-based classifier for a three-level hierarchical classification of audio signals into ten classes. In the first level, the audio signals were grouped into artificial and natural sounds. Each of the first level groups were subdivided to form the second level groups viz. instrumental, automobile, human, and nonhuman sounds. The third level was formed by subdividing the four groups of the second level into the final ten groups (drums, flute, piano, aircraft, helicopter, male, female, animals, birds and insects). A database of 213 audio signals were used in this study and an average classification accuracy of 83% for the first level (113 artificial and 100 natural sounds), 92% for the second level (73 instrumental and 40 automobile sounds; 40 human and 60 nonhuman sounds), and 89% for the third level (27 drums, 15 flute, and 31 piano sounds; 23 aircraft and 17 helicopter sounds; 20 male and 20 female speech; 20 animals, 20 birds and 20 insects sounds) were achieved. In addition to the above, a separate classification was also performed combining the LDB features with the mel-frequency cepstral coefficients. The average classification accuracies achieved using the combined features were 91% for the first level, 99% for the second level, and 95% for the third level

Audio signal feature extraction and classification using local discriminant bases

Automatic classification of audio signals is an interesting and a challenging task. With the rapid growth of multimedia content over Internet, intelligent content-based audio and video retrieval techniques are required to perform efficient search over vast databases. Classification schemes form the basis of such content-based retrieval systems. In this paper we propose an audio classification scheme using local discriminant bases (LDB) algorithm. The audio signals were decomposed using wavelet packets and the high discriminatory nodes were selected using the LDB algorithm. Two different dissimilarity measures were used to select the LDB nodes and to extract features from them. The features were fed to a linear discriminant analysis based classifier for a six group (Rock, Classical, Country, Folk, Jazz and Pop) and a four group (Rock, Classical, Country and Folk) classifications. Overall classification accuracies as high as 77% and 88% were achieved for the six and four group classifications respectively using a database of 170 audio signals.

Detection of OFDM-CPM signals over multipath channels

A class of orthogonal frequency division multiplexing-continuous phase modulation (OFDM-CPM) signals is introduced in which binary data sequence is mapped to complex symbols using the concept of correlated phase states of a CPM signal. Canonical optimum and suboptimum multiple-symbol-observation OFDM-CPM receivers are derived. Multipath channel with AWGN is assumed. The receivers are analyzed for bit-error-rate (BER) performance in terms of highand low-SNR bounds. These bounds are illustrated as a function of parameter h, time delay and attenuation level. It is shown through numerical computation that the OFDM-CPM system outperforms a similar OFDM-PSK system in a two-ray multipath model. Moreover, an OFDM-CPM system gives the best error performance when the parameter h is 0.5 with an observation interval of 2.

Design of space-time trellis codes with multi-h CPM signals

Signal design for wireless data links modeled as quasi-static Rayleigh is approached using space-time trellis (STT) coding and power/bandwidth efficient multi-h phase-coded continuous phase modulation (CPM) signals. System configuration for achieving multiple-input multiple-output (MIMO) with arbitrary number of antennas at the transmitter and receiver is described. At the receiver, the received signal is modeled as a linear sum of Rayleigh faded multi-h CPM signals with additive white Gaussian noise (AWGN). The problem of recovering transmitted data from such a composite signal is addressed using the criterion of maximum likelihood sequence estimation (MLSE). The finite-state discrete-time properties of the received composite signal are exploited and using Euclidean distance criterion best STT codes are determined. Performance analysis is presented based on pair-wise error probabilities (PWEP) for all possible pairs of signals in signal-space. It is noted that PWEP is a function of signal matrix, hence on multi-h CPM, and channel gain matrix. The signal matrix corresponding to error events is presented as a function of multi-h CPM signal parameters. From this signal matrix, the rank and the minimum determinant (minimum product of eigenvalues) of STT coded multi-h CPM can be found. It is shown that for these STT codes coding gain of 1.9 dB can be achieved relative to identical coding and MSK

Performance of optimum and suboptimum OFDM-CPM receivers over multipath fading channels

Summary A class of orthogonal frequency division multiplexing-continuous phase modulation (OFDM-CPM) signals is introduced in which binary data sequence is mapped to complex symbols using the concept of correlated phase states of a CPM signal. Canonical optimum and suboptimum multiple-symbol-observation OFDM-CPM receivers are derived. Multipath channel with AWGN is assumed. The receivers are analyzed for bit error rate (BER) performance in terms of high- and low-SNR bounds. These bounds are illustrated as a function of parameter h, time delay and attenuation level. It is shown that OFDM-CPM systems, with h ¼ 0:5; 0:25 and an observation interval length of two symbols, can outperform conventional OFDM-PSK system for a two-ray multipath model. Copyright # 2005 John Wiley & Sons, Ltd.

Generalized asymmetric multi-h phase-coded modulation for M-ary data transmission

In this paper, we describe a class of continuous phase modulation (CPM) signals referred to as generalized asymmetric multi-h phase-coded modulation (GAMHPM). The proposed signaling employs asymmetric modulation parameters, M-ary data format, and memory based past data symbols. A generalized method for choosing the modulation parameter during an arbitrary interval is presented. For specific realizations of GAMHPM signals, we present Euclidean distances and compare them with distances of well-known CPM signals. The performance estimates of GAMHPM signals in terms of probability of bit error versus signal-to-noise ratio are given.

Design and performance of an OFDM-CPM receiver for wireless communications

A new class of orthogonal frequency division multiplexing-continuous phase modulation (OFDM-CPM) signals has been introduced recently that outperforms conventional OFDM-PSK signals. These signals employ multi symbol observations for detection of data. An optimum OFDM-CPM receiver was designed that observed n symbols of the received waveform, and arrived at an optimum estimate of the data transmitted in the first symbol interval. However, from the viewpoint of receiver complexity, this receiver is quite complex and the complexity grows exponentially as a function of number of observation symbols. Thus, in this paper, we develop and investigate a less complex OFDM-CPM receiver that employs post-FFT signal processing for recovery of data. Performance of this receiver is evaluated over multipath fading channels for several subclasses of OFDM-CPM signals. It is shown that we achieve error rate performance that is far superior to the performance offered by conventional OFDM-PSK systems

Phase coded nonlinear asymmetric chirp modulation

In this paper we present a class of signals referred to as multi-mode asymmetric continuous phase chirp modulation (MA-CPCM). In these signals modulation parameters are time-varying and function of present and past data bits entering the modulator. The proposed class of signals offers additional degrees of freedom and flexibility to the designer. It is shown that constructions of MA-CPCM exit with superior distance properties compared to conventional CPCM. An addition of 1-bit memory in MA-CPCM results in a coding gain of nearly 2 dB relative to CPCM. The detection of MA-CPCM using multiple-symbol observation window is addressed and the optimum minimum-error-probability receiver is derived. It is shown that this optimum receiver offers nearly 1 dB gain over optimum CPCM receivers. MA-CPCM finds possible applications in cordless systems, wireless local area networks, air-to-ground communication via satellites, and, in general, in hostile EM environments