Sengin Posted June 19, 2009 Share Posted June 19, 2009 I'm writing a program to generate sounds from sine waves, and I have information for the harmonics, amplitude, and phase of the main orchestral instruments (tuba, trumpet, flute, oboe, etc...). I have been successful in creating a way to use the ADSR envelope in the generation of the sounds, but I am lacking information on the actual ADSR values of the instruments. I have tried searching around in various places without many results. Does anyone have this information or know where I could get some? I'm also looking for information on the spectra and ADSR of common percussive instruments (bass drum, snare, tom, cymbals, etc...) as well because I plan on attempting to generate some percussive sounding sine waves and I need some frequency data to try to replicate (though that's going to be difficult and probably not sound too much like it). If anyone has information on this I would appreciate it as well as results are even less than the instruments ADSR envelopes I've been searching for. Thanks. Quote Link to comment Share on other sites More sharing options...
zircon Posted June 19, 2009 Share Posted June 19, 2009 Unfortunately it is pretty variable for every instrument. It's hard to determine specific times. Percussive instruments tend to be easiest since they all basically have an instantaneous attack (the full tone resonates as soon as the instrument is struck.) But release times are very variable. Hearing an instrument in person is different than hearing it recorded, and even the same instrument type can have many construction types within it. My best suggestion is to base your data off recorded instrument samples you already have. Render them to WAV if they're normally in a plugin format and then analyze the wav. How long does it take for the wave to hit its highest volume? That's attack. How long does it take for it to decay to nothing? That's release. Quote Link to comment Share on other sites More sharing options...
Zephyr Posted June 19, 2009 Share Posted June 19, 2009 This shouldn't be too difficult to just do by ear, it should be easy to figure out the basic shape, then just tune it to perfection. Ex. piano would have instant attack with no sustain, just a really long decay, flute could change depending on articulation, but a medium attack with a reasonably high sustain, etc. Quote Link to comment Share on other sites More sharing options...
Sengin Posted June 19, 2009 Author Share Posted June 19, 2009 I've certainly thought about trying to do it myself and by ear (and had planned on it if I couldn't find any info), but it would be helpful and would save much time if I had the data before hand. This way, I wouldn't have to find a sample in wave format, perform an FFT to get the frequency data and try to replicate it from there. Sure, that doesn't sound so bad, but it takes a little to set up and graph the FFT as well as having to do it for every note/hit that I want (e.g. 88 times on a piano, different spots on a cymbal, etc...) - and that's even if I can find the samples I want. I was just wondering if anyone had any information so I wouldn't have to do an insane amount of work. Quote Link to comment Share on other sites More sharing options...
zircon Posted June 19, 2009 Share Posted June 19, 2009 Why do you need to set up an FFT for each note? You can use the same decay for C2 that you do for C6, pretty much regardless of the instrument. Quote Link to comment Share on other sites More sharing options...
Yoozer Posted June 19, 2009 Share Posted June 19, 2009 Does anyone have this information or know where I could get some? In other words, you're making an additive synthesizer. A lot of those have a resynthesizer, too - simply analyze the envelope for each harmonic, and then try to build a best-fit envelope curve. ADSR alone might be too lacking - a multipoint envelope would probably be of more help. Cameleon 5000 and Morphine can both do this. And like zircon says, since you can play the instrument in several ways, it's different every time. Quote Link to comment Share on other sites More sharing options...
analoq Posted June 19, 2009 Share Posted June 19, 2009 I plan on attempting to generate some percussive sounding sine waves and I need some frequency data to try to replicate (though that's going to be difficult and probably not sound too much like it). If anyone has information on this I would appreciate it as well as results are even less than the instruments ADSR envelopes I've been searching for. There's a reason for that. Notice the instruments in the SHARC collection have something in common: all those instruments produce fairly static waveforms. Instruments in the saxophone family, the piano and certain percussive instruments have changing waveforms that don't lend themselves to a representative FFT analysis. Cymbals, for example, produce a range of changing inharmonic overtones; that's why a noise generator is used to approximate them in subtractive synthesis. Quote Link to comment Share on other sites More sharing options...
Sengin Posted June 20, 2009 Author Share Posted June 20, 2009 Why do you need to set up an FFT for each note? You can use the same decay for C2 that you do for C6, pretty much regardless of the instrument. I would need to do an FFT for each note to get information about what harmonics are there are how loud they are, not for ADSR. Sorry, I should have been more clear. Yoozer: I'm looking to create a simple program to create songs from scratch, and as such I won't be using another program. I just want to be able to add something to my resume that I enjoy working on (plus I'd like to create a few songs this way). Analoq: I know the reason SHARC doesn't have any percussive instruments, and I know that percussive instruments do not have harmonics and are mostly just noise in various ranges of frequencies with a fast attack and quick decay (for the most part). I know that it is a long shot to try to recreate percussive instruments, but I'd like to give it a shot, even though it won't sound exactly like the real deal (obviously). Quote Link to comment Share on other sites More sharing options...
analoq Posted June 20, 2009 Share Posted June 20, 2009 I know that percussive instruments do not have harmonics and are mostly just noise in various ranges of frequencies with a fast attack and quick decay (for the most part). I know that it is a long shot to try to recreate percussive instruments, but I'd like to give it a shot, even though it won't sound exactly like the real deal (obviously). That's what confuses me. If you understand that the data you're looking for is useless, why do you expect that someone would bother to host and maintain it? If you want it just to play with then you need to DIY. Quote Link to comment Share on other sites More sharing options...
Yoozer Posted June 20, 2009 Share Posted June 20, 2009 I just want to be able to add something to my resume that I enjoy working on (plus I'd like to create a few songs this way). Then you could try to tackle the resynthesizer/analyzer in something like Csound, which could spew out a list of the harmonics you are supposed to use and approximations to the envelope. Educational for sure . Quote Link to comment Share on other sites More sharing options...
Sengin Posted June 20, 2009 Author Share Posted June 20, 2009 That's what confuses me. If you understand that the data you're looking for is useless, why do you expect that someone would bother to host and maintain it? If you want it just to play with then you need to DIY. The data isn't useless as it can be used for info about EQ and other effects you can add to a song. I was wondering if the FFT data for various percussive instruments was already out there. I've found some snare spectra, but that's it already (the people were comparing different snare batter heads and seeing what the spectra looked like). Then you could try to tackle the resynthesizer/analyzer in something like Csound, which could spew out a list of the harmonics you are supposed to use and approximations to the envelope. Educational for sure . Hmmm..I've heard of Csound, but I didn't think it could do that. I should look into this. Thanks! Quote Link to comment Share on other sites More sharing options...
analoq Posted June 21, 2009 Share Posted June 21, 2009 The data isn't useless as it can be used for info about EQ and other effects you can add to a song. It is useless for the purpose you (initially) wanted it for. As for equalization, practical spectral details for common instruments can be found in many recording arts books and guides online. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.