View Single Post
Old 10-24-2015, 01:45 PM
George Frankly's Avatar
George Frankly George Frankly is offline
Join Date: Aug 2011
Location: CO
Favorite beer: Sam Smith Oatmeal
Posts: 10,453

Originally Posted by Incredible Hulctuary View Post
But if you're a real audiophile and you read further about it, you'd realize how not-so-good CD audio really is.

Although the mathematical principles behind the CD audio format are sound (no pun intended), in practice things don't go so well because the theoretical concepts involved are too computationally expensive to fully implement.

The first shortfall in practice compared to the theory is the brick wall filter that is supposed to remove all frequencies above the Nyquist cutoff of 22.05kHz. Without removing those frequencies, one loses the ability to accurately reconstruct the continuous waveform from the discrete 44,100 samples per second.

But in reality there is no such thing as a true brick wall filter, i.e. completely removing all frequencies above the cutoff while leaving all frequencies below it untouched. A practical filter is going to introduce some distortion below the cutoff while still leaving something above the cutoff. The human hearing limit of 20kHz is not far below the chosen Nyquist frequency of 22.05kHz, so some filter-induced distortion can still leak into the human hearing range.

The second (and bigger) problem is that digital-to-analog converters (DACs) in consumer hardware don't even use the proven math for reconstructing the continuous waveforms, because it's too computationally expensive. Mathematically accurate reconstruction of the waveform would involve sinc functions and Fourier transforms and other math I don't understand, but that's too much for a typical $10 DAC chip to do in real time so they resort to cheap interpolations instead.

The good news is that improvements in technology are enabling the CD format to produce results closer to its theoretical ideal. In the early days of CDs, the brick wall filters were analog, and they introduced lots of distortion into the audible range so CDs back then sounded harsh. But now they're using digital brick wall filters which can perform much closer to the ideal, and they can keep improving those filters by throwing more computational power at them and tweaking the algorithms.

On the playback end of it, some high-end DACs are incorporating field-programmable gate arrays (FPGA) so they can be more true to the math that is involved with reconstructing the analog waveform. And there is software like HQPlayer that utilizes the processing power of multi-core PCs to perform that math to upsample to high rates like 192kHz or 384kHz, thereby leaving the DAC with less work to do with filling each gap between samples (provided your DAC can accept a sample rate that high, of course).
I'm not an expert in signal processing, I do agree with some of what you said and can't refute the bits I don't understand! I haven't found analog sources to sound better than CDs, so it's not like I have a better alternative.

I do want a new CD player, mine skips at high SPL. The subs shake it a lot.

I do generally say to put the money in the speakers. Distortion for sources and amps is almost academic, 0.05% or whatever. Speakers have much more. My JBL setup is at about 0.4% THD, subs are often 5% at volume. I generally focus on the drivers, the polar response, and the room. IMO.
Reply With Quote
Page generated in 0.10486 seconds with 9 queries