interesting stuff on A to D and D to A

[mylesgm]

I found this an interesting read but don't have the bandwidth here in Swaziland to watch the video...

http://wiki.xiph.org/Videos/Digital_Show_and_Tell

[Wiz]

well it took me about 30 minutes to get the video to play... 8)


AWESOME!!!!



I say that again



AWESOME!!!


man, should be compulsory watching,then you should have to pass a written test to be able to post on newsgroups.

Imagine, how much bandwitdth would have been saved..


great find!!!


Also, reminded me of my early twenties, I used to calibrate and repair that very model of sig genny and spectrum analyser....

[wez]

http://wiki.xiph.org/Videos/Digital_Show_and_Tell

I like this guy.

[Linear]

This is interesting and he does a nice (albeit slightly condescending) explanation.

There's only a few places where I reckon it falls down - sort of where you need to chuck out the test gear and objectively use your ears.

Firstly I agree with the staircase argument for waveforms. This is relevant to sine waves, because they are so even and uniform. It is also easy to see where the waveform is supposed to be, because it is repetitive, predictable and easily interpolated.

Complex waveforms however I think are a different matter. If you look at the waveform of a piece of recorded music, it isn't uniform and I definitely think that the more samples there are (ie higher sample rates) the more 'accurate' it will be. Of course going 8-bit on a sine wave is going to sound fine - however 8 bit on a passage of music sounds like ass as we all know.

And just because we can't 'hear' anything above 20kHz, doesn't mean it's not important. The harmonics that go along with the fundamental is almost just as important, and rolling everything off with a very steep filter is why we perceive that digital nastiness. Drums always sound like they've lost something in the transfer, and I reckon it's the ultra-high frequencies of the cymbals that are cut off. You can't hear them in isolation, but can perceive them when they're together with the frequencies you can hear.

Thirdly I don't agree that the 'most high end tape machines' are equivalent to 13 bit. Maybe with a sine wave and test equipment, yes. But there is no way on earth an MM1200 2" 16 track is 13 bit - even on the noise front. Tape hiss definitely sounds different to 13 bit noise - I don't know the exact reason but digital white noise is headache material.

The biggest problem I have with 'test equipment' digital arguments is that while it may be scientifically accurate, it doesn't translate to real world empirical results.

I agree that 16 bit/44kHz is fine for playback on most systems, but I think 96kHz does sound better (bit depth doesn't so much matter because most music has such little dynamic range).

Chris

[zenelectro]

There's only a few places where I reckon it falls down - sort of where you need to chuck out the test gear and objectively use your ears.

Complex waveforms however I think are a different matter. If you look at the waveform of a piece of recorded music, it isn't uniform and I definitely think that the more samples there are (ie higher sample rates) the more 'accurate' it will be. Of course going 8-bit on a sine wave is going to sound fine - however 8 bit on a passage of music sounds like ass as we all know.

And just because we can't 'hear' anything above 20kHz, doesn't mean it's not important. The harmonics that go along with the fundamental is almost just as important, and rolling everything off with a very steep filter is why we perceive that digital nastiness. Drums always sound like they've lost something in the transfer, and I reckon it's the ultra-high frequencies of the cymbals that are cut off. You can't hear them in isolation, but can perceive them when they're together with the frequencies you can hear.

Thirdly I don't agree that the 'most high end tape machines' are equivalent to 13 bit. Maybe with a sine wave and test equipment, yes. But there is no way on earth an MM1200 2" 16 track is 13 bit - even on the noise front. Tape hiss definitely sounds different to 13 bit noise - I don't know the exact reason but digital white noise is headache material.

The biggest problem I have with 'test equipment' digital arguments is that while it may be scientifically accurate, it doesn't translate to real world empirical results.

I agree that 16 bit/44kHz is fine for playback on most systems, but I think 96kHz does sound better (bit depth doesn't so much matter because most music has such little dynamic range).

Chris

Chris,

See if you can get through this video. Sorry, it's long and involved but these guys (ESS) currently make the
highest performance AD and DA chips on the market and are very involved with the subjective side of
digital audio.

http://www.youtube.com/watch?v=1CkyrDIGzOE

WRT to your reference to music versus sine waves, this is addressed directly after 40.0

cheers