skinner@saber.UUCP (Robert Skinner) (01/14/86)
Can anyone direct me to information on the harmonic content of various
musical instruments? Machine readable form would be most useful, but
I'll use any reference books you may point out. I am not only
interested in the steady state harmonics, but also the leading and
trailing edge (attack and decay) harmonics. I'm not attempting to
reproduce the sounds electronicly (yet), so the leading and trailing
harmonics are not as important now.
I'm not a regular reader of net.music or net.music.synth, so please e-mail
me any information that would be useful.
------------------------------------------------------------------------------
"Pass me the Rap Rod, Plate Captain"
"What?"
"Pass me the telephone, waiter.
If you guys were any less hip, your bums would fall off."
Name: Robert Skinner
Snail: Saber Technology, 2381 Bering Drive, San Jose, California 95131
AT&T: (408) 945-0518, or 945-9600 (mesg. only)
UUCP: ...{decvax,ucbvax}!decwrl!saber!skinner
...{amd,ihnp4,ittvax}!saber!skinnerbeaucham@uiucuxc.CSO.UIUC.EDU (01/22/86)
Re:harmonics wanted-- You want what everyone wants and for sure exists but doesn't seem to be published anywhere: a complete set of data for the harmonic (or inharmonic) spectra of all of the common acoustical musical instruments. The problem seems to be that there is an enormous amount of data to deal with. Wendy Carlos discussed this problem and her solution in working with the Crumar Synthesizer (an additive synthesis machine) in a recent issue of Keyboard (unfortunately, I can't find my copy). I think an excellent overview article on the problem of harmonic spectrum analysis of musical instrument tones is: J.C. Risset & D.L. Wessel, "Exploration of Timbre by Analysis and Synthesis" in The Psychology of Music, edited by Diana Deutsch, Academic Press (1982). Here are some articles which give some data of interest: Woodwinds: W. Strong & M. Clark, "Synthesis of Wind-Instrument Tones", JASA, Vol. 41, pp. 39-52 (1967). N.H. Fletcher, "Acoustical Correlates of Flute Performance Technique", JASA, Vol. 27, No.1, 233-237 (1975). J.A. Moorer & J. Grey, "Lexicon of Analyzed Tones: Clarinet and Oboe", CMJ, Vol. 1, No. 3 pp. 12-29 (1977). Brass: D. Luce & M. Clark, "Physical Correlates of Brass-Instrument Tones", JASA, Vol. 42, 1232-1243 (1967). J.A. Moorer & J. Grey, "Lexicon of Analyzed Tones: The Trumpet", CMJ, Vol. 1, 23-31 (1978). Strings: H. Fletcher et al, "Quality of Piano Tones", JASA, Vol. 34, 749-761 (1962). H. Fletcher & L.C. Sanders, "Quality of Violin Vibrato Tones", JASA, Vol. 41, pp. 749-761 (1967). J. Beauchamp, "Time-Variant Spectra of Violin Tones", JASA, Vol. 56, 995-1004 (1974). J.A. Moorer & J. Grey, "Lexicon of Analyzed Tones: A Violin Tone", CMJ, Vol. 1, No. 2, 39-45 (1977). I hope this helps. The basic problem is that there is an enormous amount of data to keep track of. Good luck! Jim Beauchamp ihnp4!uiucdcs!uiucuxc!beaucham
beaucham@uiucuxc.CSO.UIUC.EDU (01/22/86)
Re: harmonic info response clarification on references: JASA = Journal of the Acoustical Society of America CMJ = Computer Music Journal
rlr@pyuxd.UUCP (Rich Rosen) (01/26/86)
> Re:harmonics wanted-- > You want what everyone wants and for sure exists but doesn't seem to be > published anywhere: a complete set of data for the harmonic (or inharmonic) > spectra of all of the common acoustical musical instruments. The problem > seems to be that there is an enormous amount of data to deal with. > Wendy Carlos discussed this problem and her solution in working with the > Crumar Synthesizer (an additive synthesis machine) in a recent issue of > Keyboard (unfortunately, I can't find my copy). Not to shatter anyone's bubble, but it turns out real accurate reproductive music synthesis is not dependent on just knowing a fixed set of harmonic spectra for specific instruments. Envelope is probably at least as important as harmonic content (if not more so) in recognition of specific sounds. The harmonic spectrum (AND the envelope) often change drastically over the range of an instrument. For instance, the piano, which has a slightly inharmonic spectrum, is deliberately tuned such that octaves are "stretched" to give it its characteristic sound. (The specifics of piano tuning are a bit beyond my understanding, so if someone thinks it necessary to expand upon and correct my simplified summary, please do so.) Plus harmonic content changes very drastically over the course of a note. (This is supposed to be one of the plusses of FM synthesis. The way in which envelope control of modulation index affects changes in frequency spectrum makes for some interesting sounds.) In addition, inflection and intonation cause subtle changes in overall sound. One thing I've found from playing with the sax sample on my Mirage is that it sounds awful if you don't play it right. Played straight it sounds like a cheap organ. Played with a little inflection (the note that begins each passage started at just above the actual pitch of the note quickly bending back into pitch, plus slow variable introduction of vibrato on held notes) adds a lot of believability to the sound. I would think that precise accuracy in harmonic spectra for many instrument simulations is actually less important than many of these other factors, though in fact part of the reason an instrument like the piano is so hard to reproduce is its slight inharmonicity that defines its sound. I've been told some of the hardware in the Kurzweil is specifically devoted to idiosyncrasies of the piano and is in fact unavailable for other user defined sampling reproductions. -- Life is complex. It has real and imaginary parts. Rich Rosen ihnp4!pyuxd!rlr
beaucham@uiucuxc.CSO.UIUC.EDU (01/28/86)
>music synthesis is not dependent on just knowing a fixed set of harmonic >spectra for specific instruments. Envelope is probably at least as important >as harmonic content (if not more so) in recognition of specific sounds. >The harmonic spectrum (AND the envelope) often change drastically over the >range of an instrument. For sure, envelope and the ability to change spectra as functions of time and pitch are very important. In a performance situation it is a must for these parameters to be responsive to the player's input. However, the data one wants for instrument replication depends on type(s) and scope of synthesis he intends to use. There happen to be a lot of machines "out there" which will allow the user to give harmonic amplitudes for a fixed waveform and to specify a simple envelope. The sound that comes out of these machines will probably be no better than that of an organ imitation of an acoustical instrument. This is what I mean about the incredible amount of information that is needed. The "Lexicon" type information from the Computer Music Journal assumes additive synthesis, where each harmonic gets a completely separate envelope. If properly synthesized, this information should lead to some very realistic sounding tones--in isolation. It still doesn't say what to do over a wide pitch range and at different dynamics or articulations, but it does give a starting point. Jim Beauchamp inhp4!uiucdcs!uiucuxc!beaucham