Help With Science Fair Project

[crdto7]

I am doing my high school science fair project on the differences between different types of acoustic guitar strings. The only problem is, I know very little about how to evaluate those differences. I originally planned to use GarageBand to test the the sustain of the note (open G string), but then I discovered Audacity, which seemed much more in-depth. However, I know next to nothing about wavelengths, frequencies, etc., so I am kind of stuck. I guess my question is: How could I measure the effect different types of guitar strings have on a guitar's tone, sustain, or frequency?

[kozikowski]

Is it too late to change your project? I doubt even with all the Audacity tools, you'd be able to tell the difference between strings -- assuming groups. You wouldn't be able to tell between different steel strings or between different gut strings, but you can tell the difference between steel and gut.

The trick is to record the same guitar exactly the same way and then pluck the strings exactly the same way. Any differences would be because of the personality of the strings. I make that sound so easy. If you really wanted to go nuts, you'd make a machine to pluck the strings so the differences in your hand plucking wouldn't affect the test. Remember you have to have a reasonable quality microphone and it has to be exactly the same distance from the guitar each time and in a quiet room.

Given all that, you can do a spectrum analysis and display something like this...

http://kozco.com/tech/audacity/piano_G1.jpg

That's one piano note. The pitch of the note is down on the left somewhere, and all the rest of those spikes are overtones and they'll be different between different pianos. You should get a serious difference between steel and gut. Then you can make up some story about what each group of spikes sounds like and walk away with an A.

Koz

[kozikowski]

Oh and you need to do all this in Audacity 1.3.12. Audacity 1.2 is very old and no longer supported, updated or patched.

http://audacityteam.org/download/

Koz

[billw58]

I know next to nothing about wavelengths, frequencies, etc., so I am kind of stuck. I guess my question is: How could I measure the effect different types of guitar strings have on a guitar's tone, sustain, or frequency?

Try this wikipedia article on wavelength http://en.wikipedia.org/wiki/Wavelength This is basic physics.

Assuming the strings are tuned the same, there will be no difference in frequency between various types of strings.

By "tone" you are probably referring the mix of overtones (harmonics). As Koz pointed out, how and where you pluck the string will make a huge difference here.
Analyze > Plot Spectrum will give you a start. You can export the results of the spectrum analysis to a tab-delimited text file then import that into a spreadsheet. You can do that for multiple experiments, bring all the data into one spreadsheet then use the spreadsheet functions to compare the levels of the harmonics in the different experiments. Then use the spreadsheet to make plots.
Another complication is that the mix of harmonics will vary over time.
You'll probably want to deal with "open" strings only. The way you press your finger behind the fret will make a difference to the "tone". Or use a capo.
It will make a difference if the other strings on the guitar are muted or not. If they are not muted they will resonate in sympathy with the plucked string, contributing their own frequencies to the "tone".

"Sustain" is easy to measure - the time from when the string is plucked to where the sound decays below a specified level relative to the loudest level.

As Koz implied, this is a very ambitious project. Good luck.

-- Bill

[kozikowski]

Again if you wanted to go completely nuts (and who doesn't), take all the strings from the guitar except one. Assume steel. Then restring the guitar with one gut string tuned to the same note as the steel. That should eliminate many of the physical variations and time duration problems. Make them close like two middle strings next to each other.

Put a piece of Play Doh...
http://www.instructables.com/id/How-to- ... -Play-doh/

...or modeling clay on one string somewhere in the middle to stop it from playing and play the free one. Then reverse the clay and playing. Those are the two sound clips for analysis. You can do a slate at the beginning by speaking the date and which string is being played.

"Thursday, February 10th, Steel String" [twang twang twang]
"Thursday, February 10th, Gut String" [twang twang twang]

Koz

[crdto7]

Hopefully this isn't a double post.

Thanks a ton for the help. I probably should clarify my experiment. I have five different string types: EXP-coated bronze, steel, 80/20 bronze, phosphor bronze, and coated titanium. The string gauges and tension are the same on all. I am going to only test the open G string for each type of string by connecting my acoustic-electric guitar to my computer via a USB to 1/4in audio cable, which will eliminate any other sounds beside my guitar, then plucking each string as hard as I can directly over the sound hole. I will pluck each type of G-string at least 10 times to get some consistency. The settings on my guitar and on my computer will remain the same throughout the tests, and the strings will be as perfectly in tune as I can get them.

It is definitely too late to change my project; it's due Friday. I am using 1.3 now, as you saw from my other post. I hadn't even considered sympathetic vibrations, so thanks for that advice. Instead of using playdoh, though, I am using some washcloths. After putting those around the other strings, I noticed a huge difference in the frequency analysis. Instead of having a ton of random spikes, the only spikes I have are the overtones for G. What would differences in those overtones in different types of strings mean though? Would that explain why one string might sound mellow, and another might sound bright?

Bill- I thought sustain would be easy to measure, but I am having trouble. How can I accurately measure how loud the sound is? I can see how big the waves are, but I don't know how to measure when the crest is under a specific point. What I might end up doing is just recording for 10 seconds for each type, and then comparing how high the crest is between the different strings.

Again, thank you so much for the help.

[billw58]

Yes, the differences in harmonics could account for a "mellow" versus "bright" sound.

To measure sustain:
Select the entire pluck and do Effect > Amplify with default parameters. This will amplify the "crest" to 0 dB.
Click on the track drop down menu in the Track Control Panel and select "Waveform dB" - you can now read (more or less) the level of the sound in dB.
It might help to maximize the Audacity window and then do View > Fit Vertically (with one track in the window).
At some point the note will decay into the noise. Choose a number (-60 dB is reasonable) and click where the note decays to that level. Call that the sustain time.

-- Bill

[kozikowski]

I bet you run into the advertising froth. I bet two or more of the strings are going to sound very nearly the same and you're not going to be able to tell from the waves who's who.

Actually, that brings up another display you can do. You can edit the strings to sound next to each other as one sound file to really hear differences -- if any. That's something very few people have ever heard. You may decide in that one sound file that you really like one string type way better than the others.

Koz

[crdto7]

I bet you run into the advertising froth.

Not sure what you mean by that. As for the second sentence, I will save each sound clip with the type of string and the trial number as the title, so I won't run into any issues there.

Actually, that brings up another display you can do. You can edit the strings to sound next to each other as one sound file to really hear differences -- if any. That's something very few people have ever heard. You may decide in that one sound file that you really like one string type way better than the others.

That's a great idea and I'm definitely going to do that for my presentation. I'm also going to use computer generated tones and compare those to the guitar strings.

Bill- The problem is that I'm dealing with hundreds of waves per second, and they aren't entirely consistent. One might go under -60db, but the next couple could be above -60db or there could be a series of several in a row that are right at -60db. I think I'm just going to do what I mentioned earlier and set the first wave to 0 seconds and measure the sound at a certain point, like 10 seconds. Thanks, though.

[kozikowski]

I bet you run into the advertising froth.
Not sure what you mean by that.

Advertising if full of stories of a product not being changed at all except by something like changing the colour of the box or something else like that. People will swear up and down that one product performs better than the other. There is a TED show where a cereal maker "created" a new product by rotating their square flakes of corn 45 degrees. "New Diamond Corn Snaps."

Test groups always preferred one strongly over the other. If you do double blind tests, however, the preference vanishes.

Koz