Yes, I know that the factors you mention affect sound quality a lot. But your conclusion is falling short. I have connected my volumio system to my low- to lower middle ranking Hifisystem, and it does sound substantially less spacious and precise than my CD player. That said, I use the same Amp, same speakers, same listening position, and the same recording, played from CD and versus a bit-perfect ripped FLAC. It’s not about the volume as we played both sources sufficiently loud and there was no audible level difference. Several other listeners confirmed my impression.
Here I have to contradict again. Indeed, I do have an ESS9023 DAC (the asymmetric USB version from Hifimediy) and I am not happy with the sound reproduction as described above. I am using a Yamaha Stereo Amp, and up to now I did not really have an issue with the impedance. Moreover, it definitely has sufficient power to blow up my neighborhood volume-wise. So the ESS9023 chip alone doesn’t make everything fine, at least from my USB experience. Thus, I am much more willing to believe the other posts that vote for any I2S DACs, but it would be great if anybody can confirm that by own listening experience.
Can’t confirm in any way, that CD- players have the potential to dominate computer audio nowadays.
But! There are so many facts to be aware of! How do you rip your CD’s to get high quality files?
If you do that with 24 or more times of speed you will definitely loose the game. You’ll loose quality in a clearly noticable amount.
CD’s should be ripped with maximum 4 times speed- and that to my opinion and experience is too much!
Also you should use a serious CD ripper like EAC or db poweramp!
Don’t play any lossy file format like mp3 or similar fakes (webradio excepted). Go minimum with flac- I would prefer wav! It’s the bestsounding file format without any doubt (lag-na, I’m expecting your comment) .
Next thing is how you power your pi or whatever. With a 5$ handy PSU you won’t get the goal. Try linear power supplies
with minimum double the power (Ampere) your computing platform needs max. Use separate power supply for DAC.
Try to eliminate (deactivate- if possible) every feature that you don’t really need on a player. Don’t stream audio files via WLan- that sounds unsatisfying if you compare it with direct connected USB- or better SSD- Drive. NAS- systems are definitely not my way to go- kilometers of cables will harm sound quality.
Use high quality USB-, coax- or toslink cables (lag-na will laugh again ) but these facts I’ve tested a thousand times.
Keep I2S connections as short as possible! Try different dac’s.
Only one real life example: I’ve compared an Ayon Audio CD- Player (highly recommended by the press) in the range of 4000 € with a RPi- G2 Labs Berry NOS 1543 setup: The G2Labs- RPi combi clearly beats the player! We are talking about 4000€ vs maybe max. 300€ including all parts, housing and good quality inbuilt psu!
Take this as an inspiration. Just play with the material, spend a bit of money and don’t give up! You will succeed in the end !
Well, I remember an article where a guy played in SPDIF (or other digital output) a song and then recorded it back from that SPDIF. Then he took the recorded audio file and played and recorded it again. It’s like looping multiple times. I think he did it more than 100 times.
At the end he said he (a professional, not anyone) could barely hear the difference. I wonder how bad can the sound become after few metres of your digital link… don’t buy coat hangers and you’re fine.
And concerning NAS: my computer at this time has sent 12 GB over the local network since rebooting and there were 0 (zero) transmission errors. You may think that network cables impact audio quality, but where is your evidence? your ear (as anyone’s ear) is proven to fail and to hear what it wants, but the data are sent to the audio output from the local computer, not from the network cable, meaning that, as long as there are no transmission errors (something I basically never see), what the computer gets is what has been sent and everything before it has no impact.
If you want to argue about quality of audio card DAC ok, but network… it’s funny and maybe also sad
It’s like a book: you take it from the library and it may be worn out, with paper which turned yellowish. But as long as the characters are still readable, saying that the status of the book affects the content, is something that defies reason IMHO.
And the network test I mentioned shows that network packets ARE readable without mistakes.
With USB someone posted a link that shows that a constant stream of data is required in some transmission modes, therefore lost packets (let’s suppose that some are lost, still debatable with non-indecent cables) could really affect playback. But network? and with buffers on both sides? really… don’t complicate your life
Your arguments are interesting and aren’t in lack of realism at all.
But- i tell you only one example: recently I tried a LMS distro as a server and separated from that the squeezelite player on a Rpi. I played music files via wlan from LMS to squeezelite. The result was some kind of harshness in the highs and upper mids, reduced resolution and an unprecise rather blown up bass.
Then I drove LMS server combined with squeezelite on one raspi with an usb- connected hdd- same files-. Sound quality was increasing clearly. I really want to encourage you to try this comparison. Bass was powerful and precise now, high/mids harshness was gone and overall resolution was fine. If your theory is right, there shouldn’t be any differences between the methods. I think you mix up simple data transmission with time related data transfer that has to be carefully prepared for da conversion. If I download files over wlan I am usually missing not one bit. But with music playback it is a different thing. Make the test.
Greetingz, Robert
In addition: I think, that the critical point has to be searched between the digital source and the da converter. You can have 100% bit perfect data on this way- but if the digital data is degraded with noise and uncontrolled time conditions, a proper da conversion will fail and at this point the analog result is not perfect then. If noise could be filtered on that way and there would be precise reclocking, the result would be probably fine. But we don’t have this conditions with rpi and cheap dac’s.
As a conclusion: Why do you think that much efforts are made with the volumio- project so that it can be called an ‘audiophile music player’ if any ‘bit perfect’ player would do the job satisfying?
I use Volumio thanks to its good interface and ease of use, moreover I am more confident that the kernel is set up to actually give bit-perfect output, while other systems may perform unwanted processing.
I asked multiple times michelangelo to list the changes he applied to the kernel to be able to judge better but I never got any answer.
Back on topic, to the extent of my knowledge, there is no such thing as “the digital data is degraded with noise and uncontrolled time conditions”. Any source of this (in the digital world) counter-intuitive effect?
If the data is transmitted and received correctly and sent to the audio buffer correctly, there is no timing and noise issue anymore, only the DAC timing is left. Your quoted statement clashes with basically all the experience I can find anywhere about digital world. And clashes also on the most basic verification: if it were true and timing and noise could affect the transmitted data (even after a “correct” reception, I mean), computers would not work.
Moreover, in this specific case, if there were an issue about noise, given the compressed nature of sound, you would hear chirps and pops and other very evident distorsions.
When you mention vague concepts as you did
you are actually killing the possibility of critics/verification, because you are making a non reproducible/verifiable statements, being it totally subjective
given the subjectivity I will definitely think it depends on the fact that hearing is very strongly connected to the brain: audio signals from the ear are heavily interpreted before getting to the “conscious part of the brain”.
Also, modern technology is FAR more accurate than human senses, therefore anything that cannot be measured, basically does not exists in my opinion.
Proving statements like yours is doable, but no one ever does it. I wonder why
Maybe there has something been misunderstood. Ok, my english is far away from being perfect. As this discussion seems to lead into a dead end only some last questions:
You don’t believe that jitter (timingerrors) has an effect on sound quality? This is meanwhile a scientific fact.
Do you have an evidence?
You don’t think, that noise in the digital signal path, that is induced through bad board layouts, mains supply, bad psu’s and computer processing activity affects da processing?
Do you have an evidence?
I assume you think, that a hifiberry DAC delivers the same sound quality as for instance
an Auralic Vega DAC because the Hifiberry was measured as perfect by Daniel.
Further I think you’ll never compare the DAC’S because you are shure that your ears and your brain fool you.
Do you think, that high quality cable producer only justify their existence through successfully fooling people that all love to spend their money?
You mentioned somewhere above that people often just want to hear differences- and so they think to hear them. Can it be that you just don’t want to hear differences when you see no plausible explanation for facts?
Did you make the little setup test I recommended to you?
Btw: I absolutely don’t think that even modern measurement instruments are more sensitive than our senses. Neither I think that science is as far to have faced all the influencing facts in audio broadcasting.
I don’t complicate my live with tweaking experiments, I enjoy it because I trust my ears and brain.
I found articles about sensitivity to jitter magnitudes bigger than what we are talking about: microseconds instead of nanoseconds.
Board layout and noise can ruin the data. But it either 0 or 1: either the data is correct, or it is not. If it gets to the audio buffer without damages, whatever is behind is transparent. And it it gets to the final audio buffer corrupted, hearing wider or harsher sound must mean the corruption has been very carefully applied to the least significant bits, otherwise the effects of corruption would much more prominent.
Comparison between DACs is difficult, but I would compare them instrumentally and not by ear, for example because with the instruments I can see which one is less distorted compared to the original, with my ears I can maybe notice they are different, and that one is more pleasing, but that doesn’t mean better fidelity. You can compare with your ear how pleasant they are, but I am not sure about fidelity.
Cables have a quality plateau and some cables are better than others. The cheapest and the better ones are not the same.
But products like these are frauds: totaldac.com/ethernet_cable-eng.htm
It may be, this means we have two options. How to find out?
Instrumental measurements I would say.
the one about connecting wifi or cable? I have no wifi in the R-Pi, I use wlan only for smartphones and tablets.
But I don’t understand this part: “I think you mix up simple data transmission with time related data transfer that has to be carefully prepared for da conversion. If I download files over wlan I am usually missing not one bit. But with music playback it is a different thing.” -> from what I know, the path of the data is:
wifi -> CPU or memory -> audio DAC
wired -> CPU or memory -> audio DAC
(I reasonably assume data cannot go directly from network to audio, since it has to be taken out from network packets, processed for routing/codec, …)
If I understood correctly your theory, you are saying that the timing of the data getting to the DAC (or generic audio card) is affected be the timing and “noise” (I keep this generic term, there would be too much to go into detail) present on the network side (either wired or wifi) BEFORE the CPU/memory.
How can that be? that is not (AFAIK) how computer work: the digital signal is regenerated at every stage, therefore either you get correct data at the CPU (and whatever is before becomes unimportant) or you don’t (and you have corruption, that will be usually spotted or statistically quite evident).
well, we can use the same camera sensor to take a picture of the sun and of the stars (not at the same time)… consumer-level cameras surpassed the eyes (we cannot see the see, we cannot see the faintest stars).
And we can have 192/24 audio, that means recording frequencies at which our eardrum will never vibrate (a matter of geometry and higher-order modes of vibration), and recording them with a dynamic range far greater than I would expect anyone’s ear would be able to (it means being able to discern a 10dB chirp while six 140dB jet engines are running close to it… I don’t think we can).
So yes, I think common instrumentation has surpassed human senses (in recording, because playback is another thing) and we can verify the performances of different DACs using instruments in a more reliable way.
And we don’t even need a mic, that would be a weak point of the chain.
I cannot do it, but attaching a 192/24 recording device to the output of the DAC and recording the same file first while playing over wifi and then over wired would be an option to compare.
But you must have noticed I did not intervene in the discussion concerning DACs, since I understand they are different.
I wrote only because you mentioned the thing about the network and that was really too much
.
!
