Xiaoqing Yu

Streaming audio packet loss concealment based on sinusoidal frequency estimation in MDCT domain

A Packet Loss Concealment (PLC) method is proposed for audio streaming represented in the Modified Discrete Cosine Transform (MDCT) domain. MDCT is the dominant transform of perceptual audio coding and widely used in consumer electronics. The MDCT coefficients of a lost frame are reconstructed by our PLC method, which is based on audio sinusoidal model in MDCT domain. The sinusoid-like coefficients could be predicted by sinusoidal frequency estimation, and further optimization of computation is achieved by using sign property of sinusoids. In order to compare with other approaches, both subjective and objective evaluation tests are conducted. The experimental results show that the computational complexity and memory requirement of our approach are greatly reduced, while good quality is maintained.

An audio retrieval method based on chromagram and distance metrics

In this paper, a content-based audio retrieval method is proposed, which can quickly detect and locate similar sound in audio database. We extract a chroma-based audio feature: chromagram, a variation on time-frequency distributions, which represents the spectral energy at each of 12 pitch classes. Compared with traditional feature MFCC (Mel Frequency Cesptral Coefficient), chromagram is better when using correlation distance as audio similarity measurement. Then we choose Jonathan Footes music retrieval database to do experiments and final results show that the retrieval accuracy can reach over 96.7% using chromagram as features even when the signal-to-noise ratio is 0 dB.

Audio fingerprinting based on spectral energy structure and NMF

Audio fingerprint is also named audio hashing, which is a content-based identification technique that can be used to label an unknown audio recording. Here we introduce an audio fingerprinting algorithm based on spectral energy structure and NMF(non-negative matrix factorization) algorithm. Firstly, a matrix of audio features is generated based on the spectral subband energy of audio signal. Then NMF is performed on this matrix to capture local characteristics of audio signal, which is expected to reduce the possibility of Declassification. Experimental results suggest that the proposed scheme can achieve good recognition accuracy and can be used in applications such as broadcast monitoring and audio excerpt identification by a mobile phone.

Audio feature optimization based on the PSO and attribute importance

This paper presents a novel approach to achieve optimization for the audio features in compressed domain, which is the PSO (particle swarm optimization) algorithm basing on the attribute importance criterion of rough set theory. Our method firstly extracts the attributes of audio to form the feature vectors and pre-processes these vectors, then realizes the optimization using the proposed PSO algorithm, and finally determines the optimal feature subset. The experimental results show that feature optimization not only greatly reduces the training time of classifier, but also improves the classification accuracy. The performance of the classification model developed on the optimal feature subset. It achieves effective dimensionality reduction.

Audio similarity measure based on Renyi's quadratic entropy

Considering noise interference often exists in audio processing, it is not robust enough to calculate audio similarity by using distance measure directly. In this paper, basing on Renyis quadratic entropy, a novel scheme for audio similarity measure is proposed. In our work, we extract Mel Frequency Cepstral Coefficients (MFCCs) to represent each audio, and then calculate the similarity based on the entropy of audio samples by probability density function (pdf) of MFCCs which can be estimated by Parzen window. The experimental results show that: (a) our approach has better performance than the one based on Euclidean distance in the common SNR condition, (b) our approach can achieve 94.00% matching accuracy even when the signal to noise ratio (SNR) is 0db. In addition, our algorithm also can be applied in audio retrieval and musical cluster.

A new method of infrared sensor measurement for micromouse control

International Institute of Electrical and Electronic Engineering (IEEE) holds an international micro mouse competition each year, it is a competition full of challenge. It is needed that the competitors must master hardware circuit designs, High efficiency shortest path algorithm, searching way algorithm and controlling system. Turning function is one of the serious problems in the control of micro mouse, especially to control it when it is right-angle turn. This paper describes a new method for right-angle turn of micro mouse. In this method, value margin of the black and white part of its wheel is widened and the noise margin is increased by enlarging the relative distance between black and white stripes and infrared sensor and then the measurement accuracy of the sensor for coding wheel and reliability of the measurement are greatly improved. This method has some practical value and can be realized easily. Finally the experiments are performed and analyzed to show the effectiveness of the proposed method.

A digital tone decoder based on a modified Goertzel algorithm

According to the analysis of Goertzel algorithm, this paper presents a series of methods and measures, which are from both hardware and software, to improve the performance of the system. We implemented a tone decoder with low-cost, small package, mixed-signal chips, and it provides with its advantages of high accuracy, small size, good temperature performance and high universality. This decoder is used in many communication equipments now.

Research on fast music classification based on SVM in compressed domain

In this paper we make a research on svm algorithm used for music genre classification in mp3 compressed domain and analyze parameters combinationpsilas impact on music classification accuracy, number of support vectors and classification time. Based on the analysis, we propose a new method how to realize fast and effective music classification by svm in compressed domain. This method greatly improves the speed while the classification accuracy is guaranteed.

PEAQ compatible audio quality estimation using computational auditory model

This paper proposed an improved objective audio quality estimation system compatible with PEAQ (Perceptual Evaluation of Audio Quality). Based on the computational auditory model, we used a novel psychoacoustic model to assess the quality of highly impaired audio. We also applied the robust linear MOA (Least-squares Weight Vector algorithm) and MinmaxMOA (Minmax-Optimized MOV Selection algorithm) to cognitive model of the estimation system. Compared to the PEAQ advanced version, the proposed estimation system has a considerable improvement in performance both in terms of the correlation and MSE (Mean Square Error). By combining the computational auditory model and PEAQ, our estimation system can be applied to the quality assessment of highly impaired audio.

Audio Quality Assessment Improvement via Circular and Flexible Overlap

This paper proposed an improved audio quality metric via circular and flexible overlap. Based on the Power Spectrum Estimation via circular overlap, we use a novel circular overlap sub-frame to assess highly impaired audio. The relationship between the fraction of overlap and the metric of audio quality is also examined, and it has been proven that the accuracy of audio quality assessment increased with the fraction of overlap. By integrating circular and flexible overlap into ITU-R BS.1387, which is also known as Perceptual Evaluation of Audio Quality, our method can be applied to the quality assessment of highly impaired audio.