Music playback from hard drive vs. system memory

[gevorg]

Is there a difference in playing back music from hard drive vs. system memory?

The following paper argues that even though both ways of playback are bit-perfect (i.e. same 1s and 0s), the playback from memory is better due to lower jitter.

http://imageevent.com/cics/cmp

The paper goes further into modding the computer to reduce EMI/RFI to lower jitter even more, but lets just focus on whether music playback from system memory gives less jitter than from the hard drive.

Note that the function of a computer in this case is only storage and digital output to receiver/amplifier/DAC/etc (no Digital-to-Analog conversion inside computer).

 

[Rubycon]

Audiophile snobbery!

During musical playback the ambient noise as well as the noise floor in the recording will mask any perceivable differences. If jitter is high enough to become audible you have other problems outside of your playback medium itself!

 

[gevorg]

Let audiophiles argue whether the differences are audible or not.

The question is why would one have more jitter than the other? (from CS/engineering point of view)

 

[Mark R]

When data is retrieved from hard drive, the hard drive has increased power requirements when the heads move. This will cause a slight voltage sag in the power rails - which could theoretically cause timing differences in a PC's clock circuits. Similarly, as the hard drive sends a burst of data to the mobo, this will produce a temporary burst of RFI as the south bridge and SATA controllers jump into action - theoretically, this interference could degrade the stabilty of the clocks.

All this is moot anyway, as most decent DACs will have a data buffer with clock recovery (indeed, this is necessary with clockless digital interfaces such as SPDIF), where any incoming jitter gets averaged out over a period of a number of samples.

 

[soccerballtux]

Originally posted by: Mark R
When data is retrieved from hard drive, the hard drive has increased power requirements when the heads move. This will cause a slight voltage sag in the power rails - which could theoretically cause timing differences in a PC's clock circuits. Similarly, as the hard drive sends a burst of data to the mobo, this will produce a temporary burst of RFI as the south bridge and SATA controllers jump into action - theoretically, this interference could degrade the stabilty of the clocks.

All this is moot anyway, as most decent DACs will have a data buffer with clock recovery (indeed, this is necessary with clockless digital interfaces such as SPDIF), where any incoming jitter gets averaged out over a period of a number of samples.

Heh, on a 300W PSU there is not going to be a dip in voltage while the harddrive reads data. The harddrive consumes at most 10w.

 

[bobsmith1492]

It doesn't matter either way. Any audio data is buffered in RAM even if it is being streamed "From the hard drive." The same goes for CD players or DVD players.

Data is read in chunks from the storage medium into a RAM buffer and then fed out to the DAC at a constant rate. It's the only feasible way of streaming audio from a mechanical storage medium.

I just built a wireless audio streaming system where the audio is encoded into Ogg Vorbis format on a VLSI VS1053B, transmitted, received, buffered, and played back on another VS1053B. Without a RAM buffer, any interruption in the stream results in glitches. Plus, chunks of audio data are required for the compression and decompression processes, so a buffer is inherently required.

Originally posted by: Mark R
When data is retrieved from hard drive, the hard drive has increased power requirements when the heads move. This will cause a slight voltage sag in the power rails - which could theoretically cause timing differences in a PC's clock circuits. Similarly, as the hard drive sends a burst of data to the mobo, this will produce a temporary burst of RFI as the south bridge and SATA controllers jump into action - theoretically, this interference could degrade the stabilty of the clocks.

This might be the only possible concern. In that case, it would be best to remove the CD drive, monitor, fans, and use as slow a CPU as possible to reduce electrical noise! Then, pack the case full of capacitors on each power line... at any rate, as with everything, there comes a point of diminishing returns.

 

[spikespiegal]

If you're running to an out-board DAC, then issues with EMI/static/blah/blah/blah are irrelevant. While electrical interference inside a PC can cause havoc with onboard sound cards and such, the source of this is on the analog side. If you're running digital out, then it's a non-issue. Music data is no more prone to 'jitter' corruption then processing a spreadsheet or running a database.

Audiophiles for years have been perpetuating the myth about jitter and timing when in fact the brutal truth is that it's purely a symptom of the DAC and has little to do with the transport. A good DAC has good jitter control and buffering. The rest is mostly BS designed to sell exotic digital cables soaked in liquid nitrogen or something.

All this is moot anyway, as most decent DACs will have a data buffer with clock recovery

What he said.