(EC2015) 2015 9 - -,a),b),c),,..,suffix tree Frequency-Based Key Component Extraction -Automatic Generation of Instruction Scores for Music Video Games- Kagawa Toshimune,a) Tezuka Hiroshi,b) Inaba Mari,c) Abstract: The music video game is a game that a player plays small synthesizing device following a given instruction score to make a good ensemble with the background music which a computer plays. It is well known that generating good instruction scores is difficult. The goal of our research is automatic generation of good instruction scores. In this paper we propose frequency-based key component extraction which automatically extracts important key components from the music. Here, we assume that the phrases which occur repeatedly are important, and using suffix tree, we count the frequencies of phrases for melody, harmony and rhythm. Using this statistics, we extract the player s part, and performed user s testing. 1... 1 UT, Bunkyo, Tokyo 113 8654, Japan a) kagawa.toshimune@ci.i.u-tokyo.ac.jp b) tezuka.hiroshi@ci.i.u-tokyo.ac.jp c) mary@ci.i.u-tokyo.ac.jp. BPM 90 120 c 2015 Information Processing Society of Japan 1 326
. S T T [1][15] suffix tree 1 2 3 3 1 3 BGM MIDI 2 3 4 5 2. 2.1 beat gather[10] Any 1 BANANA suffix tree[9] Fig. 1 Suffix tree of BANANA [9]. Beats[11] [12] The Real DJ[13] beat gather Any Beats 2.2 suffix tree suffix tree Weiner S [5] [8] 1 2.3 Jekovec suffix tree [2] Jekovec MIDI [4] Ioannis aggressive accession [3] 3. 3.1 2 c 2015 Information Processing Society of Japan 2 327
2 Fig. 2 Automatic generation system of instruction scores. 3.2 beatmaniaiidx * 1 BME Be Music Script BMS [16] 2 MIDI A B BME A B 2 ( 1 ) A B ( 2 ) A B BMS Lunatic Rave2 beat2 100201 [14] 7 2 9 4 1 9 2 3 2 4 2 *1 http://www.konami.jp/bemani/bm2dx/ 3.3 BGM BGM frequencybased key component extraction FBKCE FBKCE 4 suffix tree 4. FBKCE 4.1 MIDI MIDI MIDI MIDI 1 suffix tree FBKCE MIDI MIDI C++ 2000 c 2015 Information Processing Society of Japan 3 328
Table 1 1 1 Comparison table of note infomation1. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 3-1, 3- Fig. 3 Frequency of partial musical note sequence -hanon1, 3-. 4.2 4.2.1 suffix tree 60 *2 BWV578 *3 2 1 4 2 1 22 suffix tree 1 suffix tree 1 1 *2 http://tukinoyakata.otogirisou.com/top.html *3 http://www1.ocn.ne.jp/~bach/index0.html 4-14, 20- Fig. 4 Frequency of partial musical note sequence -small fugue14, 20-. 1 2 2 3 1 3 [6] c 2015 Information Processing Society of Japan 4 329
Fig. 5 GarageBand-small fugue20-. 8 - A- Fig. 8 Frequency of partial musical note sequence -Romancing SaGa A-. 5 GarageBand - 20-6 -20-7 Fig. 6 Frequency of partial musical note sequence -small fugue20-. Fig. 7 Partially repeated note sequence 2 2 Table 2 Comparison table of note infomation. A B C 4 14 20 1 20 14 29 4 (29,4) 5 20 x 1 100 y 1 100 6 7 (ABCD...) k n (ABCD...) k 1 1 4.2.2 4.2.1 9 GarageBand - A- Fig. 9 GarageBand-Romancing SaGa A-. 2 suffix tree A B C 2 *4 13 MIDI GarageBand * 5 9 11 13 GarageBand 8 (11,5) 1 GarageBand 9 1 11 5 10 (5,5) 1 GarageBand *4 http://aika.joo.lt/ *5 https://www.apple.com/jp/mac/garageband/ c 2015 Information Processing Society of Japan 5 330
10 - B- Fig. 10 Frequency of partial musical note sequence -Romancing SaGa B-. 12 - C- Fig. 12 Frequency of partial musical note sequence -Romancing SaGa C-. 11 GarageBand - B- Fig. 11 GarageBand-Romancing SaGa B-. 11 5 5 12 (19,9) 1 GarageBand 13 19 9 4.2.3 7 5 3 1 16 8 GarageBand 4.2.2 13 GarageBand - C- Fig. 13 GarageBand-Romancing SaGa C-. 1. FBKCE 2.FBKCE + FBKCE 3.FBKCE + 3 1 2 3 2 3 14 10 5 3 [1][6] 14 c 2015 Information Processing Society of Japan 6 331
3 2 3 5. 14 FBKCE Fig. 14 Range of extraction for frequency-based key component extraction. 3 Table 3 User s testing. 1. 2(0) 2. + 7(3) 3. + 7(5) 3 7 5 14 10 5 2 6,296 1 2,640 2 2,631 3 2,637 4 50 2 30 3 1 2 3 1. 2 3 FBKCE FBKCE 2. 3. 2 FBKCE FBKCE FBKCE 2 FBKCE 3.2 [17] FBKCE FBKCE [7] [1] :,, pp 87-127 156-178 (1971). [2] Jekovec, M., J. Demsar, and A. Brodnik : computeraided-melodic-analysis-using-suffix-tree, ICMA, University of Ljubljana (2012). [3] Ioannis Karydis, Alexandros Nanopoulos, Yannis Manolopoulos: Finding maximum-length repeating patterns in music databases Multimedia Tools and Applications, Volume 32, Issue 1, pp 49-71(2007) c 2015 Information Processing Society of Japan 7 332
[4], : MIDI,. 1 1 39, pp 107-109 (2006). [5] P. Weiner: Linear pattern matching algorithms SWAT 73 Proceedings of the 14th Annual Symposium on Switching and Automata Theory, pp 1-11 (1973) [6] :,, pp 61-64, 115 (2007) [7] Diana L. Graf, BS, Lauren V. Pratt, MS, Casey N. Hester, MD, Kevin R. Short, PhD : Playing Active Video Games Increases Energy Expenditure in Children, PE- DIATRICS Vol. 124 No. 2 August 1, pp 534-540(2009) [8] http://www.hgc.jp/ tshibuya/classes/index.html (2015.04.30). [9] Jekovec, M., J. Demsar, and A. Brodnik COMPUTER AIDED MELODIC ANALYSIS USING SUFFIX TREE, http://lusy.fri.uni-lj.si/sites/lusy.fri.unilj.si/files/publications/jekovec2012-icmc12-article.pdf (2012). [10] : beat gather, http://www.konami.jp/products/touch beatgat her/ (2011). [11] : Any Beats, http://www.anybeats.jp/ (2011). [12] :, http://www.jp.playstation.com/software/title/jp 9000npjg00001 00syotenbeat000001.html (2008). [13] Float 32 Inc. : The Real DJ, https://itunes.apple.com/jp/app/the-real-djrhythm-game/id467707681?mt=8 (2011). [14] lavalse, cyclia, SHiNKA, Kamiyu, mur : LunaticRave2, http://www.lr2.sakura.ne.jp/index2.html (2008). [15] wikipedia :, https://ja.wikipedia.org/wiki/%e9%9f%b3% E6%A5%BD [16] wikipedia : BMS ( ), https://ja.wikipedia.org/wiki/bms (%E9%9F% B3%E6%A5%BD%E3%82%B2%E3%83%BC%E3%83% A0) [17] : TexTage[ ], http://textage.cc/ (2004) c 2015 Information Processing Society of Japan 8 333