Thoughts on the future of HiFi

[theaudiohobby]

I never said there was only one paper on the subject, I said I read it a couple of years ago, and I saw the test being performed at AES. I am only aware of one study, but I am pretty sure more has been done on the subject, just not for the same reasons.

In otherwords, you have precisely zero evidence to support of what is supposedly a well-accepted theory, or proven fact as you put it. Digital encoding schemes are at the heart of telecommunications the world over, if digital induced agitation were an issue, it would be widely known and published.

The DAC turns the binary data in a analog signal, and the reconstruction filter is used to construct a smooth analog signal from the output of a DAC. The question is does the waveform after reconstruction look exactly like the waveform from an all analog signal that never get's digitally processed. The answer to that is no.

My original proposition was the output of a DAC is an analog signal, a continuous waveform by definition. Your claim that the analog signal at the DAC output is not exactly identical to the original analog signal is a strawman.

Actually the issues are not so similar. Quantization noise is more audibly disturbing than the noise-floor in analog system. Dither helps can make that noise less audible, but dither can add grain to the sound as well.

At -96dB in an undithered 16-bit system across the entire passband and greater than -100dB in a dithered one, I think not. tape noise in analog tape was so bad, it required equalisation (i.e. Dolby) to mitigate it's effects.

Aliasing is indeed an issue, and if it wasn't an issue, the techniques such as oversampling and upsampling would not be necessary. Reconstruction filters that do not effect the frequency response in the higher frequencies have not been made. You have two choices, begin rolling off the signal before Nyquist frequency is reached, or create a brick wall filter which stands a good chance of ringing. Neither is a perfect solution, which is where oversampling comes into play.

not sure what your point is here, anti-aliasing is an integral part of DAC design, and like any other engineering endeavour there are trade-offs.

We can argue digital audio 101 till the cows come home but there is this one basic point. My point is simply this, I can record a live concert using split feeds from my board to two separate paths. One can be sent via a clean all analog path to an analog recorder, the other to a digital recorder or hard drive at 16bits. In a perfect world they should sound exactly alike during playback, but in reality they don't. In the end they both end up as analog, but the sonic character is clearly different.

There are any number of reasons why the scenario you describe above can happen, however none of these reasons invalidate the basic fact that the output of a DAC is an analog signal that closely resembles the encoded analog signal within the limits of practical digital design and theory. Secondly, an analog audio signal irrespective of it's source, be it a DAC or a pure analog component is always a continuous signal.