Totally agree with edwardsean. Here is my basic understanding of digital signal:
Perfect digital signal means perfect square waveform, which I have never seen in any real world scenario. How computers and networks work perfectly are heavily relied on error correction and in pure digital form (doesn't need to covert to analog). In audio, first of all, there is no error correction in s/pdif or UAC (but there are some special cases. According to Rob Watts, the Chord Windows driver will resend faulty packet. So, Hugo2's USB can have error correction). Second, the imperfect of the digital signal (shark fin waveform) can affect the timing in the D/A conversion. Third, any noise generated from the digital side can affect the RFI/EMI sensitivity analog components.
At the end, I just trust my ears. If I hear a different, go with the way I prefer. If I don't hear a different, that is great, and I am not going to worry about it.
This discussion has me in mind of experiences I have had in the distant past...
Firstly, when I first went from vinyl (on a Michel Gyrodeck) to CD the player I bought was a Sony CDP555ES. This was a very well put together device and quite weighty. In particular, the transport mechanism was an all-metal affair with a magnetic guide rail for the laser pickup. It had two transformers hanging out of the back, one was the power supply for the transport mechanism and the other was for the DAC and analogue output circuitry. The two power supplies were entirely independent.
In the early days of cd the "bits are bits" view was predominant. Demonstrations usually focused on the fact that you could damage a CD in a way that would totally destroy vinyl and the CD would still play.
It didn't take long for the high end audio manufacturers to cotton on that improving construction values in their products and making the transports more along the lines of turntables with suspension and vibration isolation led to significant sound improvements.
The fact was that vibrations picked up by the transport led to read errors. The read errors activated error correction which created digital noise in the DAC circuits. This in turn degraded the final sound quality. Hence the steps taken by Sony in the 555ES.
The DAC itself in the Sony was no great shakes and I soon moved on to using the Sony as a transport with external DACS. The first of these was the Musical Fidelity Digilog. This was a revelation at the time because, amongst other things, the silences in the music sounded so "black." It was the sonic equivalent of looking at diamonds first on beige velvet and then on black velvet. I realised then how deleterious even low-level noise is to the audio experience. The stunningly low noise floor in the DAVE is a major factor in why it sounds so good.
Later I went to a memorial mason and had a slab of black granite 1.25" thick cut to fit the Sony. I was inspired to do this by the fact that the casework of the dCS kit was so good, and in particular that the Elgar had a slab of marble embedded in its top surface. At that time I did not yet have the Verdi transport. I used Blutack to fix the granite to the TOP of the Sony and secured yet another sound improvement. This was a noticeable stabilisation of the bass. A previous slight "wandering/wavering" or "floating" sensation was gone and the bass felt more locked in.
I can only suppose that strengthening the casework of the Sony led to less vibration of the transport mechanism.
Vibration has a bad effect on all electronics, hence the plethora of after-market mechanical isolation devices and is also why high-end manufacturers like Chord go in hard on very solid casework. Even the little Mojo has solid casework.
Secondly, and years later I learnt a valuable lesson in the necessity of proper handling of digital signals from experiences with dCS kit.
I had the Verdi transport, Purcell upsampler, and Elgar DAC already in place and was auditioning the Verona digital clock. The Verona (about £4000 at the time) is a highly stable clock signal generator which, when connected to the other dCS devices, forces digital synchronisation between them. Clocking variations are a prominent source of errors and therefore noise in a digital chain. The Verona is designed to eliminate these errors.
The first track I used to test the setup was a choral piece (on a Sony disc at 20 bit resolution) recorded in a church. Prior to the start of the singing there was a few seconds of silence in the church.
Before the singing even started I know I was going to have the Verona. As the silence started it felt like the space in my room opened up and I heard the size and ambient of the church. This is, I think, due to extremely low-level sounds through the full frequency spectrum now being reproduced properly instead of being lost in noise in the digital processing.
I was just stunned. Totally unexpected.
It makes me understand why the design features and paradigm of devices like the Hugo and the DAVE result in such an amazing sonic result.
Bits are not just bits.
It is deeply important how they are managed and processed. The achievements of companies like dCS and Chord are a clear demonstration of this.
Statements made in previous posts such as the one above along the lines of ALAC sounding better than FLAC because ALAC does not involve noise associated with the digital process of unpacking the data are entirely plausible to me. I intend to make my own tests of these claims.