better musicianship through data
Feb. 28th, 2026 01:14 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
thruthelookingglassi have been having some problems with certain notes on my tenor sax. i recently got it back from the shop, and was trying to ascertain if the issues were being caused by something the repair tech did, were a result of my mouthpiece/reed combination, or something else like my technique.
Read on if you're a nerd.
I created a spreadsheet and then sat there and very tediously played every note on the horn 4 times, once with my Morgan mouthpiece and VanDoren reeds in strengths 2.5 and 3, and then repeated that with the Selmer C* mouthpiece and both reeds, using the Tonal Energy app to align to, starting at middle C and going down and then up from there. Did not attempt any unusual fingerings or anything else other than to compare bis B-flat and side B-flat. This took over an hour, to play and log all the data. As with any data collection effort, the more I did it the more I started to notice and log.
Then I popped the spreadsheet into Claude and asked it to recommend ways to improve it by adding or splitting fields etc. I hate so many things about AI, but I do feel that it in tandem with a spreadsheet is almost always really useful and genuinely an excellent tool. Can we just use it for that and not like, making deepfakes or whatever?
Then I had it help me figure out how to use and analyze the data.
I think I might need to do some comparison across sessions because I was consistently sharp on one mouthpiece and flat on the other. This is probably remedied by more careful calibration at the very beginning of the session.
I guess I can't embed an image here. Anyway it's a super useful benchmark and I'll probably do this for another couple of sessions to get more data. People are always posting the weirdest uses of AI in the forums I'm in, like using it to help make sheet music of scales (why? there are a million free ones out there, and AI usually gets it wrong).
Read on if you're a nerd.
I created a spreadsheet and then sat there and very tediously played every note on the horn 4 times, once with my Morgan mouthpiece and VanDoren reeds in strengths 2.5 and 3, and then repeated that with the Selmer C* mouthpiece and both reeds, using the Tonal Energy app to align to, starting at middle C and going down and then up from there. Did not attempt any unusual fingerings or anything else other than to compare bis B-flat and side B-flat. This took over an hour, to play and log all the data. As with any data collection effort, the more I did it the more I started to notice and log.
Then I popped the spreadsheet into Claude and asked it to recommend ways to improve it by adding or splitting fields etc. I hate so many things about AI, but I do feel that it in tandem with a spreadsheet is almost always really useful and genuinely an excellent tool. Can we just use it for that and not like, making deepfakes or whatever?
Then I had it help me figure out how to use and analyze the data.
I think I might need to do some comparison across sessions because I was consistently sharp on one mouthpiece and flat on the other. This is probably remedied by more careful calibration at the very beginning of the session.
I guess I can't embed an image here. Anyway it's a super useful benchmark and I'll probably do this for another couple of sessions to get more data. People are always posting the weirdest uses of AI in the forums I'm in, like using it to help make sheet music of scales (why? there are a million free ones out there, and AI usually gets it wrong).
