I am toallly confused as to why someone would go to extreme lenghts to setup and tweek a turntable setup for great sound and then expect to use any off the shelf Walmart PC to get high end sound. I went to great lenghts to get incredible sound from my PC. I should note that it does indeed have multiple power supplies. I got them from Granite. One on the CD ROM and one on the HD and the ATX PS for the motherboard. Nice thing is that the Granite units are retrofittable to almost any PC. They are the size of a mouse unit and easy to fit in. and it does not stop there.
Ram settings, OS tweeks and player settings all have huge impacts on digital audio.
Bits may be bits but digital audio is real time data streaming and that is the difference that opens up a big can o worms.
Ram settings, OS tweeks and player settings all have huge impacts on digital audio.
Bits may be bits but digital audio is real time data streaming and that is the difference that opens up a big can o worms.
I think all the difference lies in the PSU usage.
With low quality integrated cards you can CLEARLY (ie way higher than DAC noise) hear when the disk is working, even when it's not the one the music is being red from.
That, imho means that it has nothing to do with jitter and is related to noise coming from the HD and going into the system through the PSU.
This however does not happen with my PCI audio card (M-Audio Delta 44) listening with Sony HDR 7506 headphones.
I don't know if that is because the power is filtered in the card, or some other reason.
But i think that if an external DAC is used, all issues can be solved using an external PSU for it.
Also the idea of speed of the unit having any relevance is absurd, any HDD has at least 10 times the speed required for uncompressed PCM 24/192 7.1 audio (wich would be roughly 4.5 MB/sec, while my HDs have a sustained read speed of over 70MB/sec).
Then there are the various buffering stages, as stated by others before me: the data is stored in the ram by the player before it's sent to the sound card's buffers.
With low quality integrated cards you can CLEARLY (ie way higher than DAC noise) hear when the disk is working, even when it's not the one the music is being red from.
That, imho means that it has nothing to do with jitter and is related to noise coming from the HD and going into the system through the PSU.
This however does not happen with my PCI audio card (M-Audio Delta 44) listening with Sony HDR 7506 headphones.
I don't know if that is because the power is filtered in the card, or some other reason.
But i think that if an external DAC is used, all issues can be solved using an external PSU for it.
Also the idea of speed of the unit having any relevance is absurd, any HDD has at least 10 times the speed required for uncompressed PCM 24/192 7.1 audio (wich would be roughly 4.5 MB/sec, while my HDs have a sustained read speed of over 70MB/sec).
Then there are the various buffering stages, as stated by others before me: the data is stored in the ram by the player before it's sent to the sound card's buffers.
dht 4 me said:
Within the PC, the audio stream is split to larger chunks with relatively undemanding timing, hardly noticeable load for contemporary hardware. It is the soundcard (or USB controller) which creates the steady stream with exact timing and jitter issues. I do not believe e.g. RAM settings have any influence.
Actually a low latency system is very demanding of even a high end computer. Low latency is the standard for pro recording, mastering systems and playback. Memory settings have HUGE impacts on latency and special software is available to measure and adjust systems and memory for latency. Huge buffers do reduce cpu overhead but kill sonics and have very high latency. There is a lot to be learned from the pro audio guys. I just spent a weekend with Paul Stubblebine who does the mastering for Reference Recordings and others and learned a lot . When these things are considered the Quote of mine above has much more relevance. Interesting that guys at Pauls level hold analoge tape as the highest of standards.
Low latency undoutebly places high demand on the system performance. But WHEN do you need low latency? Definitely for recording (audio as well as midi), certainly not for sole playback (I do not mean monitoring of recording). Why should I require miliseconds latency for playing CD, flacs, streamed music from internet, etc? The pros you are talking about are in the recording business, I am talking about hifii users, i.e. doing playback.
How do buffers affect the sound apart of delays - latency?
How do buffers affect the sound apart of delays - latency?
Getting into all of the details is far beyond the scope of this blog. However there are detailed websites giving measurements and design notes out there. My office computer has them bookmarked.
FWIW Paul Stubblebine is a MASTERING ENGINEER of the highest caliber and does NOT do the recording and I mentioned all of the aspects that he talked about which includes playback. Obviously a mastering engineer who has won numerous awards should have ears and knoweledge that is of great value. Interesting this all sounds like the old "it is digital and therefore perfect" ignorance that we all once had before meaningful measurements came about. Even then there were the people who said "bits is bits". Try googling "the art of building computer transports" and see where that gets you, it is a good basic starting point.
FWIW Paul Stubblebine is a MASTERING ENGINEER of the highest caliber and does NOT do the recording and I mentioned all of the aspects that he talked about which includes playback. Obviously a mastering engineer who has won numerous awards should have ears and knoweledge that is of great value. Interesting this all sounds like the old "it is digital and therefore perfect" ignorance that we all once had before meaningful measurements came about. Even then there were the people who said "bits is bits". Try googling "the art of building computer transports" and see where that gets you, it is a good basic starting point.
I just finished skimming the Art of Building...
For a windows audio newbie it certainly offers a lot of useful tips. There are some unsupported or wrong claims, such as:
A ~2ms latency delivers performance of an equivalent $10,000 traditional CD transport. However, much lower latencies can be attained and thereby achieve far superior results. Lowest latency offers best results (~0.25ms playback latency is possible using KS)!
How does a latency relate to price of a CD player? As a matter of fact, the most expensive CD players read the disc several times using higher speed, cache the data and make corrections. They use long digital filters for upsampling which introduce very long latencies.
Another incorrect statement:
Use soundcards with non-metallic digital output interface.
Toslink is available whilst ST Glass is not easily found. For 24/96 output, Toslink excels when used with quality glass fibre cables such as Audioquest’s Optilink 5 or Van Den Hul’s Optocoupler. Galvanic isolation is achieved which is impossible with BNC, Coax or AES/EBU.
Decent PCI soundcards have their coax SPDIF output equipped with an isolation transformer and the coax is galvanically isolated from the PC circuits.
Still I have no idea how buffers damage sound and low latency is imperative to quality of the sound.
Actually, I am surprised the article does not deal with types of sound cards nor individual models which has a major impact on the resulting sound. Instead, it spends a great deal of time on tuning RAM modules timing
For a windows audio newbie it certainly offers a lot of useful tips. There are some unsupported or wrong claims, such as:
A ~2ms latency delivers performance of an equivalent $10,000 traditional CD transport. However, much lower latencies can be attained and thereby achieve far superior results. Lowest latency offers best results (~0.25ms playback latency is possible using KS)!
How does a latency relate to price of a CD player? As a matter of fact, the most expensive CD players read the disc several times using higher speed, cache the data and make corrections. They use long digital filters for upsampling which introduce very long latencies.
Another incorrect statement:
Use soundcards with non-metallic digital output interface.
Toslink is available whilst ST Glass is not easily found. For 24/96 output, Toslink excels when used with quality glass fibre cables such as Audioquest’s Optilink 5 or Van Den Hul’s Optocoupler. Galvanic isolation is achieved which is impossible with BNC, Coax or AES/EBU.
Decent PCI soundcards have their coax SPDIF output equipped with an isolation transformer and the coax is galvanically isolated from the PC circuits.
Still I have no idea how buffers damage sound and low latency is imperative to quality of the sound.
Actually, I am surprised the article does not deal with types of sound cards nor individual models which has a major impact on the resulting sound. Instead, it spends a great deal of time on tuning RAM modules timing
I also skimmed it. It recommends things like disabling the processor's L3 cache, reducing the RAM clock speed and timings. And at the end of that section, it says, "Test playback via external DAC. There should be significant improvements!" These changes affect timings by tens of nanoseconds. Their having an effect on audio playback is questionable. If there's evidence to the contrary, I'd like to see it.
On the software side, the document suggests removing as much as possible of Windows, which does make sense, I suppose, in a (no offense) obsessive-compulsive way.
If you change one tiny variable in a complex system and hear a change, color me skeptical. The underlying issue is: as long as it's got bits in it, a FIFO doesn't care where or when its bits came from.
These changes affect timings by tens of nanoseconds. Their having an effect on audio playback is questionable. If there's evidence to the contrary, I'd like to see it.On the software side, the document suggests removing as much as possible of Windows, which does make sense, I suppose, in a (no offense) obsessive-compulsive way.
If you change one tiny variable in a complex system and hear a change, color me skeptical. The underlying issue is: as long as it's got bits in it, a FIFO doesn't care where or when its bits came from.
Exactly. And the timing changes influence the input side of the FIFO (DMA plus the feedback-controlled FIFO "reclocking" from the PCI clock domain to the sound card clock domain). This kind of FIFO is there for a reason
The changes of memory timing could maybe influence noise of the power supply (which affects to some extent the sound card clock). And we are back at the power supply.
The changes of memory timing could maybe influence noise of the power supply (which affects to some extent the sound card clock). And we are back at the power supply.
I love how people spew conjecture without ever actually having tested anything. If you want to know for sure what is possible in PC Audio I suggest you start trying these things. There are scores of people now who havedone so and can validate that the methods that CICS writes about do indeed greatly improve sonics.
I suggest looking at the PC Asylum for more.
I suggest looking at the PC Asylum for more.
Wait a second, we are talking how to improve the sonics of a flawed system imho.
A PC's PSu is not made to provide power to HIFI equipment, whatever you do you can not get around that.
Instead of tweaking your system, and buying expensive SSDs, people should buy a external sound card with its own regulated PSU.
On the topic of latency, it does not affect sound quality under any aspect, it is just that when you are recording, editing, mixing or mastering you need to listen to what is under the cursor, otherwise you couldn't do a proper cut or crossfade, in instance.
When you are listening to music to enjoy it, there is no difference at all between a <10 ms and a 10 seconds system.
A PC's PSu is not made to provide power to HIFI equipment, whatever you do you can not get around that.
Instead of tweaking your system, and buying expensive SSDs, people should buy a external sound card with its own regulated PSU.
On the topic of latency, it does not affect sound quality under any aspect, it is just that when you are recording, editing, mixing or mastering you need to listen to what is under the cursor, otherwise you couldn't do a proper cut or crossfade, in instance.
When you are listening to music to enjoy it, there is no difference at all between a <10 ms and a 10 seconds system.
I use an external USB dac and I wish that latency did not have any effect on replay bit it sure as hell does especially with ASIO.
The best replay I have heard is from Wavelab and it has latency adjustments that are clearly audible.
Now I have to say that I have heard systems that were not effected by any of these settings but then again they were absolute crap in comparison to an optimized system. So yes if your system has so many other things masking the performance potential you wont benefit from any of this.
The best replay I have heard is from Wavelab and it has latency adjustments that are clearly audible.
Now I have to say that I have heard systems that were not effected by any of these settings but then again they were absolute crap in comparison to an optimized system. So yes if your system has so many other things masking the performance potential you wont benefit from any of this.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.