AES E-Library

C. So, and AN. B.. Horner, "Wavetable Matching of Inharmonic String Tones," J. Audio Eng. Soc., vol. 50, no. 1/2, pp. 46-56, (2002 January/February.). doi: C. So, and AN. B.. Horner, "Wavetable Matching of Inharmonic String Tones," J. Audio Eng. Soc., vol. 50 Issue 1/2 pp. 46-56, (2002 January/February.). doi:
Abstract: Most previous work on resynthesizing musical instrument tones with wavetable synthesis has assumed that the original tone was harmonic or nearly harmonic. This assumption does not hold for plucked string tones and bowed string tones with vibrato. A wavetable matching technique for inharmonic string tones is introduced. The technique separates partials into different classes based on their degree of inharmonicity, and uses wavetable matching for each class. Results are given for bowed violin and cello (both with vibrato), plucked Chinese pipa, and nonvibrato trumpet tones. The results indicate that the number of wavetables allocated to each class should be about 25-33% of the total number of partials in the class to achieve a good spectral match and adequate frequency deviation resolution. Listening tests have found that the new method improves the perceived match on inharmonic string tones significantly, but results in degradation on the harmonic trumpet tone compared to simple wavetable matching. This suggests that the new method is best for matching plucked string and bowed string tones with vibrato.

@article{so2002wavetable,
author={so, clifford and horner, andrew b.},
journal={journal of the audio engineering society},
title={wavetable matching of inharmonic string tones},
year={2002},
volume={50},
number={1/2},
pages={46-56},
doi={},
month={january/february},} @article{so2002wavetable,
author={so, clifford and horner, andrew b.},
journal={journal of the audio engineering society},
title={wavetable matching of inharmonic string tones},
year={2002},
volume={50},
number={1/2},
pages={46-56},
doi={},
month={january/february},
abstract={most previous work on resynthesizing musical instrument tones with wavetable synthesis has assumed that the original tone was harmonic or nearly harmonic. this assumption does not hold for plucked string tones and bowed string tones with vibrato. a wavetable matching technique for inharmonic string tones is introduced. the technique separates partials into different classes based on their degree of inharmonicity, and uses wavetable matching for each class. results are given for bowed violin and cello (both with vibrato), plucked chinese pipa, and nonvibrato trumpet tones. the results indicate that the number of wavetables allocated to each class should be about 25-33% of the total number of partials in the class to achieve a good spectral match and adequate frequency deviation resolution. listening tests have found that the new method improves the perceived match on inharmonic string tones significantly, but results in degradation on the harmonic trumpet tone compared to simple wavetable matching. this suggests that the new method is best for matching plucked string and bowed string tones with vibrato.},}

TY - paper
TI - Wavetable Matching of Inharmonic String Tones
SP - 46 EP - 56
AU - So, Clifford
AU - Horner, Andrew B.
PY - 2002
JO - Journal of the Audio Engineering Society
IS - 1/2
VO - 50
VL - 50
Y1 - January/February 2002 TY - paper
TI - Wavetable Matching of Inharmonic String Tones
SP - 46 EP - 56
AU - So, Clifford
AU - Horner, Andrew B.
PY - 2002
JO - Journal of the Audio Engineering Society
IS - 1/2
VO - 50
VL - 50
Y1 - January/February 2002
AB - Most previous work on resynthesizing musical instrument tones with wavetable synthesis has assumed that the original tone was harmonic or nearly harmonic. This assumption does not hold for plucked string tones and bowed string tones with vibrato. A wavetable matching technique for inharmonic string tones is introduced. The technique separates partials into different classes based on their degree of inharmonicity, and uses wavetable matching for each class. Results are given for bowed violin and cello (both with vibrato), plucked Chinese pipa, and nonvibrato trumpet tones. The results indicate that the number of wavetables allocated to each class should be about 25-33% of the total number of partials in the class to achieve a good spectral match and adequate frequency deviation resolution. Listening tests have found that the new method improves the perceived match on inharmonic string tones significantly, but results in degradation on the harmonic trumpet tone compared to simple wavetable matching. This suggests that the new method is best for matching plucked string and bowed string tones with vibrato.