Every now and then, I turn my focus away from radio signals and see what interesting things may be found doing spectral analysis on everyday sounds or music. This time, it is U2 music I found featuring rehearsals and sound checks conducted in 2001 for the song "New York."
During late April and early May of 2001, U2 was trying to tweak the song intro and ending. The band is known for putting a lot of work into its performances; lots of unofficial and unreleased recordings show U2 devoting time and sweat into small details. In the case of "New York," it appears that U2 wasn't satisfied with the drumming and bass, for being a bit too heavy in the low end. Also, they added a snippet of "You Only Live Twice" and had to try a few times to get the timing right in reference to some looped music as "New York" opens.
I wanted to see how Bono and The Edge's voices looked in relation to the instruments. They sound different, but do they look different under FFT spectral analysis? How do guitar notes look when The Edge runs his instruments through distortion and delay units? For the answers, I used my trusty Catbird Linux system, with a temporary installation of Audioprism.
Installing Audioprism was easy and fast, taking only a few minutes to install some dependencies and compile the binary (see Github page for details). With the software installed, I simply opened it from the command line then played some interesting audio to analyse. What you get is a new window, which I had set up to scroll sound in an FFT waterfall moving from right to left. Here is the command, if you want to try it:
audioprism -r 24000 \ --orientation horizontal \ --magnitude-min -15 \ --dft-size 2048 \ --width 1280
In the video linked above, you can see a full sequence of U2 rehearsing the song "New York." It starts with The Edge calling for a run through of "New York" and then a bit of "Elevation." I believe the comments were to Joe OHerlihy, who's reply is at a lower level in the background. Bono then appears, wanting to work on clearing a rumbly low end and improving the intro timing.
In the spectrum, you can clearly see the different voices, clean notes, and notes with warble, repeats, or grunge added in. The recordings end with a live performance of "New York" from Chicago, made in May, 2001. It is a stereo mix of Bono (vocal) and Adam Clayton (bass) in-ear-monitors. What is really slick here is the presence of whistling fans, which make unique wiggly lines in the spectrum around the top of the vocal range. Clapping and the murmur of thousands of people appear more "noise like" in the background.
Watching the spectrum of music performances, especially live shows, is a different sort of animal from watching radio signals. I can see the value in identifying unique performances based on the background sounds. Effects such as reverb, repeat, distortion can be seen and directly evaluated on the waterfall. In short, FFT for live music is a cool thing, and the coolest is watching U2 as a live spectrum.