Yes, that's totally correct - the exact same bits of data are being dumped to the DAC via the USB port. But,the processing elsewhere, beforehand, to create that stream of bits differs.
Sorry, player software very strongly affects the sound quality - I recently did a round of experimenting with an old laptop, just using the internal sound system to monitor SQ. And the precise way the software gets the job done had major impact. Why? The software I now use, Media Monkey, intelligently buffers the track, and uses close to zero cycles of CPU time to get the job done - the laptop goes to sleep while the music is playing. The standard tool, Media Player, fidgets constantly while the track is playing, you can see the CPU cycles being burned up throughout the playing. And the difference is quality is quite substantial, because the machine is generating less interference internally, during the process.
Sorry, player software very strongly affects the sound quality - I recently did a round of experimenting with an old laptop, just using the internal sound system to monitor SQ. And the precise way the software gets the job done had major impact. Why? The software I now use, Media Monkey, intelligently buffers the track, and uses close to zero cycles of CPU time to get the job done - the laptop goes to sleep while the music is playing. The standard tool, Media Player, fidgets constantly while the track is playing, you can see the CPU cycles being burned up throughout the playing. And the difference is quality is quite substantial, because the machine is generating less interference internally, during the process.
Last edited:
If the bits are identical, the sound is identical. The only thing that can get in the way is the operating system and underlying sound system layers. This is why people choose players and/or operating systems which allow for easy bypassing of all that stuff. If you hear a difference between Media Player and Monkey something, with lossless formats, then one or both of those players is doing something more than decoding the file and dumping it to the DAC. The bits are different. For if they were identical, if they were both "perfect" then they would sound identical. No amount of system activity, or lack thereof, will make the bits any more or less perfect. They are either delivered or they aren't. Machine "interference" does not alter bits of data. Errors can occur, but the protocols and architecture are designed to detect and correct the errors. Computers would be useless, otherwise.
Last edited:
No. On the Data In pin of a DAC, at one end of the trace transferring the signal to the converter chip or circuitry it's digital, at the other end it's analogue - once the signal enters converter land all bets are off, it's in a strange hybrid world. A nasty bit of noise decides to run down the supply pin of the DAC circuitry, and the internals go whoopsy doo!! - we didn't do that that last bit of converting work quite right, did we, boys?!!
It's all about, yes, that everything is still always analogue - so long as the interpretation of what's happening is, remains digital then everyone's happy, no surprises, ever! But, our ears need analogue, and that's where things start getting mucky. As soon as the 'data' starts being interpreted as having analogue meaning, the door opens for mischief to happen, and, unfortunately, it usually does, to at least a small degree.
It's all about, yes, that everything is still always analogue - so long as the interpretation of what's happening is, remains digital then everyone's happy, no surprises, ever! But, our ears need analogue, and that's where things start getting mucky. As soon as the 'data' starts being interpreted as having analogue meaning, the door opens for mischief to happen, and, unfortunately, it usually does, to at least a small degree.
And so it begins ...
(the second block of bolding is mine)
This would be caused by one thing: insufficient isolation between the digital processing portions and the D/A converter. It's not a bit surprising that this happens using a line/headphone out from a laptop.
An external USB interface CAN be a lot better, but I wouldn't bet that they all are.
Don't blame 'bad' software for bad electronic isolation. Most computers' sound outputs are the audio equivalent of an old boom-box.
benb, when I say internal sound I mean the works - so, it's using the internal speakers, those tiny nothings behind the keyboard. Of course this makes everything that much harder in terms of getting acceptable SQ, but I was curious how much it could be "improved", and what were relevant factors. Turns out the software was a major factor, because it reduced the impact of that "bad electronic isolation".
The same considerations scale up to the most ambitious, fully high end setups - time spent in "debugging" every aspect can reap real rewards.
The same considerations scale up to the most ambitious, fully high end setups - time spent in "debugging" every aspect can reap real rewards.
Decompress , buffer the uncompressed out ahead. Usually 2 separate
processes. FLAC codec uses some CPU cycles to create the uncompressed
PCM stream the output decoder sends to the sound card/DAC.
Usually the former takes place >2000ms before the latter - my buffer is 2500ms.
whether the flac is compressed or uncompressed , it still goes through these
two "stages" , what comes out is identical.
edit - the flac codec will also perform CRC (like a zipfile) to make sure the output
is a perfect lossless copy. That is why a compression level 8 flac is STILL
lossless !
OS
Last edited:
Take a wav file , zip it with winzip.
unzip it back to original ,it's bitperfect.
Programs (and windows) would crash if this did not work flawlessly.
This is all that happens on a compression level X FLAC.
My main HT PC even does a full file buffer (decompresses the whole file) ....
then plays it.
Edit - winzip is the perfect analogy , it also has different compression levels.
OS
unzip it back to original ,it's bitperfect.
Programs (and windows) would crash if this did not work flawlessly.
This is all that happens on a compression level X FLAC.
My main HT PC even does a full file buffer (decompresses the whole file) ....
then plays it.
Edit - winzip is the perfect analogy , it also has different compression levels.
OS
Last edited:
Guys, the pit of deliberate ignorance is impossible to fill. Don't even try, it just creates more gas.
Meh, people can rave all they want about the advantages of digital downloads (and argue endlessly over the formatting of such), I like to buy music in a format I can actually hold in my hands and that won't be gone forever in the event of a hard drive failure. And nothing beats actually buying that music directly from the artist's merch table at a show, it just feels so much more personal than downloading a bunch of bits of data off the internet.
These files we speak of are on one of my blu-rays (in my hand - 1K files). And
on a second blu-ray ... PLUS on the hard drive for me to listen to them.
My purchased (even old) CD's , are in a box. With perfect and MP3 copies in 3 more spots.
Some MP3's on portable player memory sticks (or phones)
I'm sure the CD's will last packed in the box ??
PS - some of my titles have 5 !!! backups ..... no losses.
OS
If you do, I'll give you 100 bugs. I swear to drunk, I'm not god. The square root of 25 is 5, but square root of -1 is i. A man a plan a canal panama. So much nonsense...
No, I'm talking about a buffer that's as long as the track. That is, the uncompression may be done, say, an hour before playing, 🙂 - the only processing is reading the data from the storage device and passing it to the DAC area, with minimal extra CPU engagement.
Yes, this is saying that CPU activity affects the sound - now isn't that strange, electrical behaviour in one area has impact in another area - tsk, what a bizarre concept!
Not bizarre, but in this case simply incorrect.
It doesn't matter what is going on inside that PC. Crappy power supply, swollen CPU supply caps, massive electromagnetic interference from a nearby light ballast, hamsters running around on the video card... You name it. It just doesn't matter, because binary data has no noise.
The signaling which transmits the data may have low noise or it may have excessive noise, but in the end the ones and zeros are noise free. There will be absolutely no difference once those ones and zeroes make their way to the DAC.
Don't fool yourself, Frank. You are doing yourself a disservice.
fas42
I have two questions:
First:
It wasn't a day ago you quoted the forum rules to a moderator. I'm struggling to understand this last comment's final line. What is your intent?
Second: In what way are you saying that a digital signal interacts with a dac in a way other than providing it the 1's and 0's that is otherwise modulated by a cpu in a moderately low power state vs. a completely idle state. In both states the signals being passed are of a digital hi/lo nature so exactly what is your point other than stating that the load of the cpu is putting draw on the the digital ground.
Because a bizarre concept is exactly what I infer, because any DAC with a reasonable output scheme will decouple digital ground from analog ground and even then have some sort of DC servo to offset any analog ground fluctuation in realtime.
Please explain in great detail with your point and exactly how I can reproduce your observation. I have a scope, tell me how I can duplicate your findings, or how you produce the error on your setup. Please provide details.
If it isn't reproducible, it does not exist. If a credible sample of listeners can't qualitatively hear the anomaly, it doesn't exist.
I have two questions:
First:
It wasn't a day ago you quoted the forum rules to a moderator. I'm struggling to understand this last comment's final line. What is your intent?
Second: In what way are you saying that a digital signal interacts with a dac in a way other than providing it the 1's and 0's that is otherwise modulated by a cpu in a moderately low power state vs. a completely idle state. In both states the signals being passed are of a digital hi/lo nature so exactly what is your point other than stating that the load of the cpu is putting draw on the the digital ground.
Because a bizarre concept is exactly what I infer, because any DAC with a reasonable output scheme will decouple digital ground from analog ground and even then have some sort of DC servo to offset any analog ground fluctuation in realtime.
Please explain in great detail with your point and exactly how I can reproduce your observation. I have a scope, tell me how I can duplicate your findings, or how you produce the error on your setup. Please provide details.
If it isn't reproducible, it does not exist. If a credible sample of listeners can't qualitatively hear the anomaly, it doesn't exist.
That's where we have the disagreement - those one and zeros as they enter the DAC are now analogue signals, and unless those one and zeros always behave precisely the same way, and all other signals inputting to the DAC, including the power supply connections are sufficiently 'perfect', from an analogue perspective, then, there may be a problem.
Binely, thanks first of all for that response - it was excellent, and I wish more people here would use such an approach! 😎
First of all, it does irritate me that common sense seems to desert the scene at times, so I express my annoyance by being a bit sarcastic - sorry about that, 😱 !
One way my point can be expressed is like this: consider the DAC area as being a true Black Box, with various inputs - consider all the power supply lines, and grounds, as being inputs! - and a key output. Electrically set this apart from everything else, and then deliberately exercise any one of the inputs by adding noise, or a specific imperfection to the signal waveform, and monitor the output. A good approach would be simultaneously feeding two instances of these Black Boxes, one remains always clean, and the other gets fed varying amounts of dirt, and do a simple diff of the outputs. Then, try combining "dirtying" of inputs as seems reasonable - at some point, with enough dirt, one must get significant results - from then on work backwards to find out how much the Black Box needs to be isolated, to ensure adequate performance.
Last edited:
Yes. Analog signals. Either electrical or optical. In this case they are a "wrapper" or "vehicle" for the ones and zeroes. The music, however, is in the ones and zeroes, not in the "wrapper". When the analog signals reach the other end, the receiver device (USB chip, SPDIF decoder, etc.) converts the signalling back to pure binary data (ones and zeroes) and then "wraps" it again into a different type of signalling (typically i2s) etc. etc.
The analog "wrapper" can change several times and can carry with it noise of all sorts, but ultimately the ones and zeroes carry no noise.
Now, analog signalling noise can be carried (electrically, but not optically) from the PC to the DAC which is where isolators come in. I'm not arguing that point, because it is a real issue.
However, if the source and DAC are isolated (either optically or otherwise) then the noise of the source is entirely irrelevant. A "busy" CPU will not produce ones and zeroes any differently than an idle one. Of course, if the CPU is so pegged with interrupts that it simply cannot put the bits into the pipe quick enough, then the sound "quality" will be plagued with extremely audible effects known as "no music", or at the very least "stuttering".
Focusing on the power supply in the DAC itself makes sense. Focusing on anything in the PC other than ensuring bit-perfect playback (which is easy enough to achieve on any operating system) is a waste of time.
Last edited:
- Status
- Not open for further replies.