口ドラムによるドラムパターン検索手法  [in Japanese] A Drum Pattern Retrieval Method by Voice Percussion  [in Japanese]

Search this Article

Author(s)

Abstract

本稿では,人がドラムの音を真似て口ずさんだ音声(口ドラム)を認識し,それに対応するドラムパターンを検索する手法を提案する.従来,実際のドラム音(楽器音)を対象とした認識は研究されてきたが,口ドラムは研究されていなかった.口ドラム認識では,声質とドラム音表現の両方の個人差への対処が問題となるため,従来のドラム音認識手法は適用できない.そこで本手法では,擬音語を中間形式として採用することでこの問題に対処する.擬音語の各音素を口ドラム音のスペクトル構造へ対応付けるために確率モデルを用い,声質の個人差を吸収する.また,各ドラム音に対応する擬音語の辞書を用意して,表現の個人差に対処する. 200発話の口ドラムデータに対して実験した結果, 91.5%の認識率を得た.

This paper proposes a method of recognizing voice percussion (simulated drum sound by voice) and retrieving the corresponding drum pattern from a database. Although drum sound recognition has been the topic of existing work, there has been no previous attempt that dealt with the problem of voice percussion recognition. This problem is difficult because of individual differences inherent in voice spectrum characteristics and also in how the intended drum sounds are articulated. We solve this problem by utilizing phonemic sequences of onomatopoeia as internal representation. The sequences are estimated from the input power spectrum with a stochastic model, and are flexibly matched with dictionary entries representing typical drum patterns. This two-level scheme is intended to deal with the two types of individual differences mentioned above. In an experiment with 200 utterances of voice percussion, our method achieved a recognition rate of 91.5%.

Journal

  • IPSJ SIG Notes

    IPSJ SIG Notes 55, 45-50, 2004-05-07

    Information Processing Society of Japan (IPSJ)

References:  13

Cited by:  1

Codes

  • NII Article ID (NAID)
    110002947079
  • NII NACSIS-CAT ID (NCID)
    AN10438388
  • Text Lang
    JPN
  • Article Type
    Journal Article
  • ISSN
    09196072
  • NDL Article ID
    6985491
  • NDL Source Classification
    ZM13(科学技術--科学技術一般--データ処理・計算機)
  • NDL Call No.
    Z14-1121
  • Data Source
    CJP  CJPref  NDL  NII-ELS 
Page Top