Wide-Band Audio Coding Based on Frequency-Domain Linear Prediction

Motlíček, Petr; Ganapathy, Sriram; Heřmanský, Hynek; Garudadri, Harinath

doi:10.1186/1687-4722-2010-856280

Cited by 3 publications

(5 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These base cepstral features with their first and second derivatives, together with audio energy and its first and second derivatives, compose the 45-dimensional feature vector. FDLP [17]: We also incorporate the Frequency Domain Linear Prediction (FDLP) feature into the system. The FDLP model, in contrast to the short-term analysis by MFCC, is based on linear prediction on different frequency bands, and describes the perceptually dominant peaks and removes the finer-scale details.…”

Section: Audio Featuresmentioning

confidence: 99%

Multimodal feature fusion for robust event detection in web videos

Vitaladevuni

Zhuang

et al. 2012

2012 IEEE Conference on Computer Vision and Pattern Recognition

146

126

View full text Add to dashboard Cite

Combining multiple low-level visual features is a proven and effective strategy for a range of computer vision tasks. However, limited attention has been paid to combining such features with information from other modalities, such as audio and videotext, for large scale analysis of web videos. In our work, we rigorously analyze and combine a large set of low-level features that capture appearance, color, motion, audio and audio-visual co-occurrence patterns in videos. We also evaluate the utility of high-level (i.e., semantic) visual information obtained from detecting scene, object, and action concepts. Further, we exploit multimodal information by analyzing available spoken and videotext content using state-of-the-art automatic speech recognition (ASR) and videotext recognition systems. We combine these diverse features using a two-step strategy employing multiple kernel learning (MKL) and late score level fusion methods. Based on the TRECVID MED 2011 evaluations for detecting 10 events in a large benchmark set of ∼45000 videos, our system showed the best performance among the 19 international teams.

show abstract

Section: Audio Featuresmentioning

confidence: 99%

Multimodal feature fusion for robust event detection in web videos

Vitaladevuni

Zhuang

et al. 2012

2012 IEEE Conference on Computer Vision and Pattern Recognition

146

126

View full text Add to dashboard Cite

show abstract

“…Another approach to estimate the minimum-phase component is to perform linear prediction in the spectral domain, 12 which is the dual of the classical time-domain linear prediction and has found good use in many speech and audio processing applications. [13][14][15] B. Group delay models and applications…”

Section: Related Literaturementioning

confidence: 99%

“…13 This spectral prediction model, also known as frequency-domain linear prediction (FDLP), allows for a convenient parametric representation of a transient. 14 The model parameters consist of regression coefficients and prediction residual. The regression coefficients capture the slowly varying temporal envelope of the signal.…”

Section: Comparison Of Segd Model With Frequency-domain Linear Prementioning

confidence: 99%

Spectral-envelope–group-delay models for transients

Shenoy

Seelamantula

2013

The Journal of the Acoustical Society of America

View full text Add to dashboard Cite

Transient signals such as plosives in speech or Castanets in audio do not have a specific modulation or periodic structure in time domain. However, in the spectral domain they exhibit a prominent modulation structure, which is a direct consequence of their narrow time localization. Based on this observation, a spectral-domain AM-FM model for transients is proposed. The spectral AM-FM model is built starting from real spectral zero-crossings. The AM and FM correspond to the spectral envelope (SE) and group delay (GD), respectively. Taking into account the modulation structure and spectral continuity, a local polynomial regression technique is proposed to estimate the GD function from the real spectral zeros. The SE is estimated based on the phase function computed from the estimated GD. Since the GD estimation is parametric, the degree of smoothness can be controlled directly. Simulation results based on synthetic transient signals generated using a beta density function are presented to analyze the noise-robustness of the SEGD model. Three specific applications are considered: (1) SEGD based modeling of Castanet sounds; (2) appropriateness of the model for transient compression; and (3) determining glottal closure instants in speech using a short-time SEGD model of the linear prediction residue.

show abstract

“…Equation 4 shows the content description vector using extracted LPC coefficients after uniting the two programme segments.…”

Section: Set the Two Description Vectors Satisfyingmentioning

confidence: 99%

“…Hotho et al [3] proposed a backward-compatible multichannel audio codec combining a high audio quality and a low parameter bit rate. Motlicek et al [4] applied frequencydomain linear prediction not only for coding speech but also well suited for coding other important acoustic signals such as music and mixed content.…”

Section: Introductionmentioning

confidence: 99%

Digital media information compression through auditory content analysis

Zhang

et al. 2010

Proceedings of the Second International Conference on Internet Multimedia Computing and Service

View full text Add to dashboard Cite

Lots of repetitions exist in television programmes and other kinds of broadcastings. Besides traditional data compression, a new information compression method is proposed to detect repeated programme segments in order to further reduce information redundancy. This approach is different from former data compression methods in that it realizes compression in semantic level through extracting auditory contents rather than signal statistic information. This is the reason for which it is named auditory content-based information compression. In our method, LPC features are extracted at acoustic level, an LPC distance measure is defined to divide digital media stream into programme segments, and auditory content description is created by statistically weighting LPC coefficients for each programme segment. Applying the idea of LZW lossless data compression, a Content-based Information Retrieval algorithm is developed to complete information comparation, selection, combination, and updating of the segments, similarly as data compression and decompression operations. Finally depending on well accepted evaluating criterion, our experiments indicate an approving information compression result and confirm the completeness of the audio information redundance detection and the reversibility of the compression process.

show abstract

Wide-Band Audio Coding Based on Frequency-Domain Linear Prediction

Cited by 3 publications

References 28 publications

Multimodal feature fusion for robust event detection in web videos

Multimodal feature fusion for robust event detection in web videos

Spectral-envelope–group-delay models for transients

Digital media information compression through auditory content analysis

Contact Info

Product

Resources

About