A Study on Music Summarization

음악요약 생성에 관한 연구

  • Kim Sung-Tak (School of Engineering, Information and Communications University) ;
  • Kim Sang-Ho (School of Engineering, Information and Communications University) ;
  • Kim Hoi-Rin (School of Engineering, Information and Communications University) ;
  • Choi Ji-Hoon (Digital Broadcasting Research Division, ETRI) ;
  • Lee Han-Kyu (Digital Broadcasting Research Division, ETRI) ;
  • Hong Jin-Woo (Digital Broadcasting Research Division, ETRI)
  • 김성탁 (한국정보통신대학교 공학부) ;
  • 김상호 (한국정보통신대학교 공학부) ;
  • 김회린 (한국정보통신대학교 공학부) ;
  • 최지훈 (한국전자통신연구원 디지털방송연구단 방송미디어연구그룹) ;
  • 이한규 (한국전자통신연구원 디지털방송연구단 방송미디어연구그룹) ;
  • 홍진우 (한국전자통신연구원 디지털방송연구단 방송미디어연구그룹)
  • Published : 2006.03.01

Abstract

Music summarization means a technique which automatically generates the most importantand representative a part or parts ill music content. The techniques of music summarization have been studied with two categories according to summary characteristics. The first one is that the repeated part is provided as music summary and the second provides the combined segments which consist of segments with different characteristics as music summary in music content In this paper, we propose and evaluate two kinds of music summarization techniques. The algorithm using multi-level vector quantization which provides a repeated part as music summary gives fixed-length music summary is evaluated by overlapping ration between hand-made repeated parts and automatically generated summary. As results, the overlapping ratios of conventional methods are 42.2% and 47.4%, but that of proposed method with fixed-length summary is 67.1%. Optimal length music summary is evaluated by the portion of overlapping between summary and repeated part which is different length according to music content and the result shows that automatically-generated summary expresses more effective part than fixed-length summary with optimal length. The cluster-based algorithm using 2-D similarity matrix and k-means algorithm provides the combined segments as music summary. In order to evaluate this algorithm, we use MOS test consisting of two questions(How many similar segments are in summarized music? How many segments are included in same structure?) and the results show good performance.

References

  1. L. Mani and M. T. Mabury, Advances in Automatic Text Summarization, Cambridge, Massashusetts: MIT Press, 1999
  2. F. Ren and Y. Sadanaga, An Automatic Extraction of Important Sentences Using Statistical Information and Structure Feature,NL 98-125, pp. 71-78, 1998
  3. Y. Gong and X. Liu, 'Summarizing Video by Minimizing Visual Content Redundancies,' IEEE International Conference on Multimedia and Expo, pp. 788-791, Tokyo, Japan, 2001
  4. I. Yahiaoui, B. Merialdo and B. Huet, 'Generating Summaries of Muli-Episode Video,' IEEE International Conference on Multimedia and Expo, pp. 792-795, Tokyo, Japan, 2001
  5. Matthew Cooper and Jonathan Foote, 'Automatic Music Summarization via Similarity Analysis,' Proc. IRCAM, pp. 81-85, 2002
  6. Jonathan Foote, 'Visualizing Music and Audio using Self-Similarity,' Proc. ACM Multimedia Conference, pp. 77-80, Orlando, Florida, November 1999
  7. Beth Logan and Stephen Chu, 'Music Summarization Using Key Phrases,' IEEE International Conference on Audio, Speech and Signal Processing, pp. 749-752, 2000
  8. Chansheng Xu, Yongwei Zhu, and Qi Tian, 'Automatic Music Summarization based on Temporal, Spectral and Cepstral Feature,' IEEE International Conference on Multimedia and Expo, pp. 117-120, 2002