Solutions
There are a couple of different strategies to get the best sound quality out of a PC.
Hardware
The claim is that a PC with components like high frequency clocks, electro motors , switched power supply, etc. is a source of RFI, EMI in such a way that it disturbs the DA conversion.
In essence it generates too much sample rate jitter in the DA conversion.
A couple of answers:
Nonsense
If there is something sensitive to ‘noise’ it is video. If your PC can generate pristine video, there is no evidence of noise whatsoever.
Nothing
Press Play and enjoy the music!
The funny thing is this might be a good strategy.
More and more you will find postings like these on audio forums
I remember that not too long ago:
- motherboard audio sounded awful
- it was almost impossible to get bit perfect audio output on a vanilla setup
- PCs were noisy
No surprise that a lot of people have gone to extreme lengths to overcome the above problems (including myself).
Nowadays:
- you can get motherboard audio chipset's that actually sound decent, and comparable to high-end DAC's (my subjective impression, and verified by Audio Rightmark measurements)
- Vista and MacOS both provide a relatively clean audio path even on WMP and iTunes (Linux as well)
- You can get completely noiseless PCs off the shelf (Dell Mini 9 is a recent example)
Given that, I think it's perfectly possible to get very decent audio with very little fiddling. Of course, it is possible to get incremental improvements by optimization and careful selection of components, but it is also possible to damage the audio experience.
Christine Tham
Optimize
A normal PC
- Remove all unnecessary programs
- Disable any unnecessary devices
- Run audio software with high priority
- Remove all unnecessary cards from the motherboard
- OS on solid state
- Use a high quality sound card
- Use a memory player
A typical and popular example is cMP², a combination of software (cPlay) and hardware (cMP, a dedicated audio PC).
Minimize
A low power PC running a minimal and efficient OS like Linux
- Headless, no screen, keyboard or mouse
- Run a program like MPD (play back server)
- Fan less, no hard disk, boot from flash drive or SSD,
- Linear power supply
- Top quality sound card or USB audio
With the recent thread on Linux breaking the ice, I thought I'd share some thoughts on an interesting use for linux, the low powered network client.
John Swenson
Maximize
The opposite is also advocated; buy a modern high power computer.
The faster it is, the more memory you add, the better the sound.
We did double blind testing in the following way;
1) 2 computers MacBook Pro and MacBook (MacBook Air at last RMAF) both dual boot Vista Ultimate/OSX with JRiver and iTunes (we did not use amarra).
2) We would play a WAV and a FLAC of the same song. We would play an ALAC and AIFF. 10 songs total... most of the time like 30-60 seconds these would be randomly mixed (i.e. sometimes FLAT PCM first [AIFF/WAV] sometimes lossless first). Each person was given paper to record the scores.
Conclusion...
a) 93% could tell the difference with the slower MacBook/Air. This is over 50 people now at more than 4 shows.
b) Using a faster machine in the same test brought that number down to 58%.
Gordon Rankin
Streaming
A streaming audio player has a small processor, no hard disk, no fan.
It is a very minimalist computer, you don't need much power to play audio.
This is in principle the ideal solution.
However, most of these devices are targeted at the consumer market.
This market is about price not about the best possible sound quality.
In practice sound quality is poor to good but never great.
You can improve on it by using an outboard DAC.
As the digital output is jittery, excellent input jitter rejection is required.
Isolate
Even the best quality clock driving the best DAC chip won't help you, as the electrical noise inside a PC is far too high to get optimum DA conversion.
Use an outboard DAC.
This not necessarily will cure the problem as you have to feed the DAC with a signal from the computer and this signal (SPDIF/Firewire/USB) might have high jitter content.
A lot of DACs, including the very expensive audiophile ones are sensitive to input jitter.
The DAC should have excellent input jitter rejection.
Buffer the input, reclock it or use ASRC (Asynchronous Sample Rate Conversion)
There a separate reclockers on the market.
One might argue that if the DAC is immune to input jitter, any input will do.
For years you heard of people removing jitter from SPDIF inputs and now Firewire and USB with the use of upsamplers or reclockers. Well actually both of these methods merely reduce the jitter they cannot remove it and the more there is the more that get's through.
J. Gordon Rankin
A solution might be to decouple the DAC as much as possible from its source.
- A protocol like asynchronous USB to eliminated input jitter (the DAC times the bus)
- Galvanic isolation of the digital input using an optical connection to avoid any stray signal entering the DAC
- Filter on the AC inlet
SD card
If all those components are a source of RFI, EMI, disturb the power, etc sure a PC is not suite as an audio source.
Just stick to your CD player.
Well, the transport has a motor, its head is moved by a linear actuator and these are again sources of disturbance!
Ok, than a SD card player is the only answer.
Software
Software jitter
You can find claims about software induced jitter.
All data inside a PC is buffered, even if software in some way or other would send the data with some sample rate jitter to the audio device, it will be buffered there.
If software induces jitter it is possibly indirectly.
If more processing power induces more jitter in the DA, an efficient media player, frugal on computer resources will sound better.
Memory player
If access to the hard disk has a negative impact on sound quality, read the song in memory before playback starts.
The advantage to memory play is reduction of activity in the computer box during playback. Normally, data is read off of the disk into memory and then copied to the device, so there are two separate activities going on. With memory playback only copying data from memory to the device is happening. Ideally, the hard drive could be spun down as well, so there wouldn't even been any electrical and acoustic noise while listening to music.
The assumption is that less activity means a cleaner signal out of the computer box to the DAC. Whether this is true or not, and whether or not this affects the output of the DAC will depend on the specific system.
The primary disadvantage to memory playback is the time spent up front loading memory. There is also an extra cost of large amounts of memory, which will be measured in gigabytes when playing long hi-res playlists. For example, I have one double album that would require 5 GB just to hold the WAV file for the complete program.
Mixer
All major operating systems have a mixer.
If you play multiple audio streams, they must be converted to the same sample rate as sound cards are not able to play at different sample rates at the same time.
If you play a single audio stream, its sample rate must match one supported by the hard ware.
Simple sound cards only play at 48 kHz. If you play a CD (44.1 kHz), sample rate conversion must be applied.
Sample rate conversion is complex, the one supplied by your operating system at its best does a fair job but not top.
K-mixer (XP) does a very bad job. Likewise older versions of OSX
A simple way to avoid sample rate conversion is to set the sample rate of the OS to the sample rate of your audio, in general 44.1 (CD).
Of course, your hardware (sound card) must support this.
Another option is to bypass the mixer.
ASIO on Win XP or WASAPI on Vista/Win7 allows for straight communication with the sound card bypassing the audio engine.
See Foobar on how to compare DS en WASAPI.
Enhancements
The audio driver often comes with all kind of bells and whistles.
Atmospheres like ‘stadium’, ‘powerful’, ‘voice cancellation’, etc.
The greatest enhancement is in general to disable all enhancements.
The world of OS
The three major operating systems on the desktop are Windows, OSX and Linux.
Getting familiar with computer audio and learning a new OS at the same time is probably only for masochists. Stick to the OS you are used to.
You will find many hot debates, most of all between Mac and Win (the Linux guys are too busy with hacking to be able to join) but there isn't any hard evidence that one is superior to the other as far as sound quality is concerned.
All are able (if configured rightly) to deliver bit perfect output.
They all run on more or less the same hardware.
Not to be mistaken for running on all hardware, that’s the stronghold of Windows.
A lite OS by itself should not affect sound quality.
Interestingly, most of the major operating systems used today on PC hardware (Windows, MacOS, Linux) are all based on micro-kernels.
Of the three, Linux is in fact the "least" micro-kernel of them. Linux basically took the Minix code base (which Andrew Tanenbaum wrote using a micro-kernel philosophy) and made it more monolithic.
Now, the good thing about micro-kernels is that the heart of the OS is very tiny - basically a pre-emptive scheduler, a shared memory mechanism, the ability to create threads or processes with very little overhead, and a simple but robust IPC mechanism.
Given that, there is no real advantage in running say a lite distribution of Linux (which comparatively has a bigger core kernel) compared to let's say a fully bloated installation of Windows. Standard XP will run happily in 512MB of memory with no paging (I explicitly disable virtual memory).
On both, you can control exactly which services, drivers etc. get loaded.
So, given that, to me the important factor for an audio PC is the quality of the drivers, and specifically what gets executed in kernel mode vs user mode.
Christine Tham
Local truth
What to think of all these different strategies.
Should you order one of those small PCs, get a linear power supply, say goodbye to a GUI, move to WAV or AIFF, etc.
I have no reasons to believe that what works in one system, will work on another system too. We talk technology and that is equivalent to a lot of parameters and each might affect different systems in a different way.
Besides, our preferences differ.
Personally I didn’t switch to computer audio to have a minimalist interface.
Browsing my collection in all kind of ways, cover art, good overview (big screen) are all part of my computer audio fun.
Part of the computer audio fun is that there are a lot of things you can easily test.
Best way to test is an ABX test; it protects you (and your bank account) against your own prejudices.
The impact of hard disk access during playback on sound quality.
You can try:
- Foobar with a huge buffer, loading a song entirely in memory
- You can also try the reverse, increase HD access during the play back by running a backup.
The impact of processor activity during playback on sound quality.
Claims like FLAC (lossless compression) sounds less than WAV (lossless without any compression) because the decompression generates more system activity can be tested by ripping the same tracks to FLAC and to WAV.
Wi-Fi
The impact of Wi-Fi can be tested by using a wired connection (see to it that the Wi-Fi card is truly disabled).
No, you can’t disable the Wi-Fi of your neighbors.
Well you can, but it won’t improve the relation.
If you want your listening room free of RFI you need a Faraday cage!
File formats
Does the fact that decoding FLAC generates more CPU activity than playing WAV have a negative impact on sound quality?
Does the fact that FLAC requires only 60% of disk I/O compared with WAV have a positive impact on sound quality?
Rip a couple of tracks to FLAC and to WAV and do an unsighted listening test.
More about WAV versus FLAC.
PSU
If you have a laptop you can test the impact of the power supply.
Plays a song with the mains connected and play it with the mains disconnected.
Now you have a clean DC and the switching power supply with all its HF content is inactive. Will this affect the sound?
It probably will, you can even measure it.
Apple MacBook Air 5G - NwAvGuy