Dissertation > Excellent graduate degree dissertation topics show

Music Feature Analysis and Its Application in Content-based Retrieval

Author: XueZhenWu
Tutor: ZhouJun
School: Shanghai Jiaotong University
Course: Signal and Information Processing
Keywords: Baseband matrix Dynamic Time Warping Highly dynamic adjustment Variable-length search
CLC: TP391.3
Type: Master's thesis
Year: 2009
Downloads: 89
Quote: 0
Read: Download Dissertation

Abstract


With the development of the Internet and the advent of the information age, the increasingly large number of digital music. Today there are many sites online music online play and download these sites are often a collection of hundreds of thousands or even more music; even on a PC music collection will usually have thousands of first. More and more music to make people richer artistic experience, but also brought a lot of difficulties to the music library management and retrieval, it is necessary to study new intelligent music management and retrieval methods. Humming-based music retrieval is an intelligent music retrieval method, it is a fundamental difference between the traditional text-based retrieval method has. Of this study is the humming-based music retrieval, it involves two key issues: how accurately extracted features from music (usually pitch extraction) and how to accurately match the characteristics (usually using dynamic Time Warping algorithm). Characteristic analysis, humming-based music retrieval vast majority are using the characteristics of the pitch as the music, so the key to the analysis is how to accurately extracted from the music pitch characteristics. Interference of background music on the pitch extraction, this paper proposes a new matrix based on the fundamental frequency of the pitch extraction algorithm, the algorithm uses the baseband matrix pitch extraction, to identify the most likely base from all possible baseband frequency value, even in a strong background music is still able to accurately extract the singer singing voice of the pitch, thus ensuring the accuracy of the retrieval system. Commonly used dynamic time warping algorithm can achieve time alignment feature matching can correct the humming input time error but the humming input in addition to the time error tone error, dynamic time warping algorithm has been improved , the introduction of the dynamic adjustment of the height (pitch), the dynamic adjustment in the characteristics of the process of matching the pitch of the humming input, and so be able to correct the humming input pitch error and improve the accuracy of the feature matching. Also taking into account the feature matching algorithm complexity is too high, this paper introduces a search of the variable length, it is after the starting point is known match time to obtain the length of the matching, simplifying the complexity of the search, thereby increasing the retrieval speed. Top 10 accuracy in this paper, and the improved algorithm of 2250 wav format music music library and 100 experimental humming input, 87% Top 70% Top 59% Top 1 has reached 36%, baseband matrix pitch extraction algorithms and improved dynamic structured algorithm is feasible and effective.

Related Dissertations

  1. ASIC Design of Specific Speaker Isolated Speech Recognition System,TN912.34
  2. Research of Voice-controlled Order System for Iatrical Equipment,TH77
  3. Study of Speaker Recognition System Based on Support Vector Machines,TN912.34
  4. Off-line Handwritten Signature Texture Feature Extraction and Verification Based on Spectrum Analysis,TP391.41
  5. Notation of Speaking Face Based on Video and Text Infomation,TP391.41
  6. Fast Time Series Similarity Matching and Its Application Research in Molten Iron Silicon Content Modeling,TF513
  7. Signature Verification Based on Video,TP391.41
  8. On-Line Handwriting Signature Verification and Its Evolutionary Algorithm Implementation,TN918.2
  9. The Recognition Model Research Based on Whole Acoustic Structure Features of Speech Unit,TN912.3
  10. Study of Mandarin Digit Speech Recognition Algorithm Based on HMM Model,TN912.34
  11. Content-based Music Information Retrieval via Humming/Singing,TP391.3
  12. Based on the combination of VQ and DTW speech recognition algorithm,TN912.34
  13. Voice Control Applications,TN912.3
  14. Application Research of Speech Evaluation in Embedded System Based on Characteristic Comparison,TN912.3
  15. Chinese speech preprocessing and isolated word recognition method,TP391.42
  16. Flight control system fault diagnosis technology research and software development,V267
  17. Online Signature Verification for Log on of Tablet PC,TP393.08
  18. Speaker Recognition System Research and Implementation,TN912.34
  19. Study of Speech Recognition System for Mandarin Digit Based on HMM,TN912.34
  20. Research & Implementation of DSP on Speaker Recognition,TN912.34

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Retrieval machine
© 2012 www.DissertationTopic.Net  Mobile