Melody Slot Machine

“Melody Slot Machine”  is an interactive music system that provides an experience of manipulating a music performance. The melodies used in the system are divided into multiple segments, and each segment has multiple variations of melodies. By turning the dials… (READ MORE)


The 15th International Symposium on Computer Music Multidisciplinary Research Theme: AI Era 2-6 November 2020 Welcome to CMMR2020!


Melody Slot Machine Web-based time-span tree editor (2020) Manual time-span generation tool (2018) Sound Scope Headphones Deep Drone Concert Viewing Headphones BandNavi ShakeGuitar Fingering Simulator 親子情報共有システム/ Wearable device for children’s safety 予測ピアノ/ Melody expectation method based on GTTM and TPS… (READ MORE)


Manual time-span generation tool (2018) 国際会議論文 [PDF] [GTTM] [分析データ・ツール] Masatoshi Hamanaka, Keiji Hirata, Satoshi Tojo: “GTTM Database and Manual Time-span Tree Generation Tool”, Proceedings of the 15th Sound and Music Computing Conference (SMC2018), pp.462-467, July 2018. [PDF] [GTTM] [分析データ・ツール] Masatoshi Hamanaka, Keiji… (READ MORE)


国際会議論文 [GTTM] [演奏支援] Nami Iino, Mayumi Shimada, Takuichi Nishimura, Hideki Takeda, Masatoshi Hamanaka: “Proposal of an Annotation Method for Integrating Musical Technique Knowledge Using a GTTM Time-Span Tree”, Proceedings of the 25th International Conference on MultiMedia Modeling (MMM2019), Lecture Notes in Computer… (READ MORE)

deep GTTM

International conference paper [PDF] [GTTM] [Deep GTTM] Masatoshi Hamanaka, Keiji Hirata, Satoshi Tojo: deepGTTM-II: Automatic Generation of Metrical Structure based on Deep Learning Technique, 13th Sound and Music Conference (SMC2016), pp.221-249, 2016. [PDF] [GTTM] [Deep GTTM] Masatoshi Hamanaka, Keiji Hirata,… (READ MORE)

Masatoshi Hamanaka

About (en)

Masatoshi Hamanaka received PhD degree from the University of Tsukuba, Japan, in 2003. He is currently a leader of Music Information Intelligence team in Center for Advanced Intelligence Project, RIKEN. His research interest is in music information technology, biomedical and unmanned aircraft systems…. (READ MORE)

Awards and articles

受賞等 [Melody Slot Machine] IJCAI-19 (the 28th International Joint Conference on Artificial Intelligence), Most Entertaining Video Award, 2019 [論文PDF] 情報処理学会論文誌 Vol.56 No.3 特選論文, 2015. [論文PDF] [創薬] 情報処理学会 第77回 大会優秀賞(Best Paper Award of IPSJ National Convention), 2015. [論文PDF] [exGTTM] ICMC2005 Best Paper Award, (Journal of New Music Research… (READ MORE)

Structures of films

References [PDF] Seiko Takeuchi, Masatoshi Hamanaka, Junichi Hoshino: “Method of structuring a film based on the Generative Theory of Tonal Music”, IPSJ Special Interest Group on Human Computer Interaction, 2015-HCI-162, No. 6, 8 pages, March 2015. [PDF] [Structures of films] [exGTTM] Seiko Takeuchi, Masatoshi Hamanaka: “Structure… (READ MORE)

Researches- Drone

我々は,救急車に先駆けて到着し,救急管制への状況の連絡や,AEDなどの携帯型の医療機器による初期治療を行う「救急ドローン」を開発し救命率を向上させることを目指している. ドローンは,道路渋滞の影響を受けず,物流や救助など様々な分野での利用が期待されているが,上空を多くのドローンが飛行するようになると,衝突の危険性が増加し,高速で飛行させることが困難になることが予想される. そこで我々は,ドローン飛行網の設計および飛行網上を飛行するドローンを管制するシステムを構築する上で重要となる,高精度な位置推定手法の構築を進めてきた.     ▲飛行網の例   ▲地表面の断面形状の取得   国際会議論文 [PDF] [ドローン] Masatoshi Hamanaka: “Deep Learning based Area Estimation for Unmanned Aircraft Systems using 3D Map”, Proceedings of 2018 International Conference on Unmanned Aircraft Systems (ICUAS2018), pp.416-423, June 2018. 国内会議論文 [PDF] [ドローン] 浜中雅俊:… (READ MORE)

Researches- Drug discovery

本稿では,医薬品となる化合物を発見するための第一段階のスクリーニングである,タンパク質と化合物の相互作用の予測について述べる.膨大な種類の化合物から医薬品になり得るリガンド化合物を見つけ出す工程は,開発にかかる時間とコストを押し上げる主要因となっている.我々はこれまで,相互作用が確認された12.5万件の結合データと,結合データに含まれない同数の組み合わせを非結合データとして用意し,それらをサポートベクターマシンで学習することで相互作用を予測するCGVBS法を提案してきたが,データが増えるにつれて学習時間が長大になることや,学習データが少数追加された場合でも再度学習をやりなおさなくてはならないなど,今後大規模な相互作用データを学習していく上で検討すべき課題があった.そこで,相互作用予測にDeep Learning(深層学習)の一手法である,Deep Belief Networks(DBN)を用いたCGBVS-DBN法を提案する.400万件のデータセットを用いた評価実験の結果,CGBVS-DBN法は98.2%の精度で相互作用が予測可能であった。 図: 論文 [LINK] [創薬] 浜中雅俊: “in silico創薬におけるスクリーニングの高速化・高精度化技術”より, “第2章 スーパーコンピュータ・人工知能による創薬・育薬の高速化と最適化”, “10節ディープラーニングを用いた化合物—タンパク質の相互作用予測”, 技術情報協会, 2018. [LINK] [創薬] Masatoshi HAMANAKA, Kei Taneishi, Hiroaki Iwata, Jun Ye, Jianguo Pei, Jinlong Hou, Yasushi Okuno: “CGBVS-DNN: Prediction of Compound-protein Interactions Based on Deep Learning”, Molecular Informatics, Vol.36, Issue1-2,… (READ MORE)

Separation of music signal sources

本稿では複数の楽器による混合音源をスペクトル包絡保存に基づく NMF によって個別の楽器の音響信号に分離する方法を提案する.我々の手法は近しい音高であればそのスペクトル包絡が周波数方向にシフトするという調波音の一般 的な特徴に基づいたものであり,各々の楽器の演奏可能なあらゆる中心周波数において基底を設け,隣接した基底についてその包絡が近似するよう制約を行うこ とで分離を行う.この操作は各々の楽器の音色を特徴として成分を分離することに相当し,教師なし学習での調波・非調波混合音源の分離を可能とする.性能評 価実験として MIDI での混合演奏音源を分離し,SNR を求めた結果リードギターについて約 3.6dB,ドラムについて 6.0dB の分解能を得ることができた. This  research  proposes  a  method  to  separate  polyphonic music signal into  signals of  each  musical instrument by NMF: Non-negative Matrix Factorization based on preservation of spectrum envelope. Sound… (READ MORE)

Music Analysis based on Implication-Realization Model

We propose a melody generation system based on the Implication-Realization Model (IRM) of music theory. The IRM is a music theory, which was proposed by Eugene Narmour. The IRM abstracts music. It then expresses music according to symbol sequences based… (READ MORE)

Guitarist Simulator: Learning-based jam session system

This paper describes a jam session system that enables a human player to interplay with virtual players which can imitate the player personality models of various human players. Previous systems have parameters that allow some alteration in the way virtual… (READ MORE)

Fingering Simulator: Guitar fingering estimation from monophony

Fingering Simulator optimizes guitar fingering of monophony by using physical simulator called Springhead2. Unlike most previous guitar fingering optimization systems focusing on moving distances, our method deals with the torque of the finger joints. Experimental results showed that the maximum… (READ MORE)

Concert Viewing Headphones

We designed concert scope headphones that are equipped with a projector, an inclination sensor on the top of the headphones, and a distance sensor on the outside right headphone. We previously developed sound scope headphones that enable users to change… (READ MORE)

Composing music with Twitter

We built “Twitracker” that can post a monophonic melody on Twitter and can create a song by placing posted melodies as materials on Twitter. A shared unfinished song is put into the compositional process by allowing modifying. Most previous composing… (READ MORE)


BandNavi Ver. 1.1 Language: Japanese, English NewForestar Co,.Ltd A band-member-backtracking application called BandNavi has been developed that enables a user to discover new songs and bands by tracing musicians who have played in different bands.Previous similarity-based song recommendation systems only… (READ MORE)

BandNavi HD

BandNavi HD Ver. 1.0 Languages: Japanese and English New Forestar Co,.Ltd BandNavi HD is an iPad version of an iPhone App called BandNavi. BandNavi HD enables a user to notice relations between musicians, and to experience a new way to… (READ MORE)

Discussion structure editor based on music theory

The discussion structure editor generates time-span trees expressing degree of importance of statements. 会議記録の自動要約には、会議の流れや重要発言の認識といった、会議記録の分析が必要です. 本研究では、音楽の分析手法であるGTTMを応用して、会議記録の分析方法の構築を行っています. 音楽においては音イベントが、会議においては発言が時間の進行とともに発生し、ゲシュタルトを生成する相似点に着目しています。 References [PDF] Hiroya Miura, Kenta Togashi, Masatoshi Hamanaka, Katashi Nagao, Satoshi Tojo, Keiji Hirata: “音楽理論を応用したディスカッションマイニングにおけるタイムスパン木と延長木の自動生成について”, IPSJ SIG Digital Contents Creation, Vol. 2013-DCC-3, No.10, January 2013…. (READ MORE)

Polyphonic Music Time-Span Tree Analyzer

We have been developing a music analysis system called a polyphonic music time-span tree analyzer (PTTA). A time-span tree assigns a hierarchy of ‘structural importance’ to the notes of a piece of music on the basis of the Generative Theory… (READ MORE)

Melody expectation method based on GTTM and TPS

A method that predicts the next notes is described for assisting musical novices to play improvisations. Melody prediction is one of the most difficult problems in musical information retrieval because composers and players may or may not create melodies that… (READ MORE)

Melody morphing method based on GTTM

ShakeGuitar (Free), based on the morphing method, is available on App Store (Released on September 26, 2009) This paper describes a melody morphing method that generates an intermediate melody between a melody and another melody with a systematic order according… (READ MORE)

σGTTM System

σGTTM combines the generative theory of tonal music (GTTM) and statistical learning. We previously devised exGTTM, which has accommodated the original GTTM to computer implementations. The exGTTM has adjustable parameters, these parameters have to be manually configured. Therefore, it is… (READ MORE)