Groove Thang | Forums

[muzac989]

To detect the note a fast fourier transform doesn't require much duration at all. This is probably what is being used, not MIDI. The initial spike isn't a whole heck of a lot different than the rest either. With the speed of a USB connection on a nice clear guitar signal a low E could probably be easily detected in 10ms. The programmers would know this latency though and "backdate" the info to align it with the music. It really doesn't seem that there should be any delay issues on what's going into the system. What's coming out of course is all over the place depending on what it's going through.

My understanding is that a FFT still requires a sizable window for the more complex sounds, and I assume they use a "one-size-fits-all" algorithm for detecting single notes and chords. Also note that the tech is capable of pulling out "correct" single notes out of chords, which probably also has implications. I'm just wondering how this effects the game scoring triggers, because the game scores you in real-time. PS: I haven't studied acoustic physics in years, and I am no programmer, so I might just be talking straight out of my a$$.

[Steel_Nirvana]

You got me thinking about this...and it's not like RS has to figure out what note you are playing; it just has to monitor for a particular frequency. So while the attack will create all sorts of strangeness, I'd think a bandpass filter of some sort could make the detection simpler. But my specialty is radar, not acoustics (and even that's been a while), so I could be way off....

[Sornborger]

You're correct. They could look for the sound level impulse or even just the level to indicate "something was done" whether that was a string being hit or a Hammer On. Other things like slides wouldn't require that trigger. Then just watch the particular frequency (ies) in question to see if they are present at a level above the other junk like harmonics. If you get Transcibe! transcription software and look at a song/note with the frequencies on it's really clear what the graphic representation of an fft looks like. It's fairly easy to detect a note with drums, bass and some other noise in there, with just a pure, undistorted guitar signal all by itself it must be a piece of cake, umm, once you've invested man years into the software and tested it extensively and had the idea in the first place.......