Hacker News Reader

Comment by laconian | Hacker News Reader

It gets way out of tune the farther you get from middle C.

Haha, yeah, because the sampling rate is super low, in order to make the data URIs work. Turn on "visual mode" hit 8 or click on the lightning bolt, and you'll see the visual representation of each sound—and you'll see how it breaks because of the sample rate when it starts sounding weird.

Looks like your sample rate is 2048Hz. Shannon-Nyquist says that should be plenty to represent the fundamental of your highest note (C5 at 523Hz). Even if you're getting aliasing on some of the higher harmonics, I'd think the tuning would still sound okay.

Any acoustics nerds that can help me understand this?

The notes are not sine waves.

The harmonics (sines whose wave length is an integer fraction of the fundamental) play a big role in how we perceive sounds.

Right. I understand what harmonics are.

I'd expect timbre to get weird from aliasing of the high harmonics, but since all the partials within an octave above that C5 are under the Nyquist frequency of a 2048Hz sample rate, I'm a little surprised that the pitch is affected.

I guess there's probably some destructive interference being caused by the aliasing though. Maybe that's what's ruining the tuning.

I just bumped up the sampling rate to 11025Hz and that seems to have helped.

The pitch sounds alright to me up until the B above middle C.

Reply on news.ycombinator.com