Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

People used to say human eyes can't perceive >60fps.

It's also just CDs, not digital formats in general. Grab an audiophile and ask their opinions about digital PDM/PCM formats, high bitrate AACs even, against true vinyls. They wouldn't have as much opinions as they do against CDs.

Also: 44.1kHz sampling rate != arbitrary waveform up to 22050Hz, unless music you're listening to consists of pure sine waves(and not even classic Yamaha FM sound chip signals).





Anybody can distinguish 60fps from higher frame rate just by looking at steady motion. The famous Blur Busters Test UFO makes it easy:

https://testufo.com/

But in the case of analog recording, nobody can distinguish a pure analog recording from the same thing but with a good ADC/DAC pair in the signal path in a blind test. It's theoretically possible to hear undithered 16 bit quantization noise if you turn the volume up extremely loud, but correctly mastered CDs should be dithered from higher bit depth.

And 44.1kHz sampling rate can theoretically represent arbitrary waveforms up to 22050Hz. The only complication is that this requires a brickwall filter, which is impossible to implement. That's why the sampling rate is set higher than needed to exceed the 20kHz limit of human hearing (in practice the limit for adult hearing is almost always lower). The higher sample rate allows for a practical filter with a shallower transition band to be used.


Isn't this true?

- 44.1ksamples/sec can only represent arbitrary waveforms at some point lower than 44.1kHz/2.

- Example: The only 22.05kHz waveform you can encode at 44.1ksamples/sec is a square wave (for 16 bit samples: -32767, 32768, -32767, 32768, etc.)

Going down to 44,099 samples/sec you could only do an extremely crude "steppy" approximation of a sine wave, sort of like the NES's triangle channel.


No, because a reconstruction filter is used to remove the stairsteps. This does not lose any information. I recommend watching the xiph.org videos explaining it:

https://wiki.xiph.org/Videos

EDIT: Also, consider that true square/triangle/sawtooth waves are mathematical abstractions that can't exist in reality. If you try to move a real loudspeaker cone in a square wave, you have to reverse direction in exactly zero time. This requires infinite acceleration and therefore infinite force. If you take the Fourier transform of these waveforms you get an infinite series of harmonics.

A real-world "square" wave only contains the lower harmonics within some frequency band. When you limit it to audio frequencies, all square waves above 6.67kHz are identical to sine waves because the only harmonic within that frequency band is the fundamental.


> Also: 44.1kHz sampling rate != arbitrary waveform up to 22050Hz, unless music you're listening to consists of pure sine waves

Every signal can be represented as a combination of pure sine waves. That insight is the basis of Fournier analysis / transform.


It's also (pretty much) how sound is processed by the inner ear. The different little hairs each pick up different frequencies.

Joseph Fourier enters the chat…



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: