MQN testing/experimentation thread

Anything to do with computer audio, hardware, software etc.
jkeny
Posts: 2387
Joined: Sun Jan 17, 2010 9:37 pm

Re: MQN testing/experimentation thread

Post by jkeny »

Forgive the double posting but I thought this audio example may well serve as a test audio files for analysis as it gives us something specific & concrete that can be heard & therefore a good candidate fo analysis.

I posted this on the MQN thread also
I find 2.66 SSE2 dqa dqa unique in that it is the only player that renders the intro to one of my test tracks correctly (at least I think it is correct).
The test track is track 1 "Court & Spark" from Herbie Hancock's "Letters to Joni".
With all other players (including previous versions of MQN) I hear a brief, varying, low level hiss noise after the strike of each piano chord in the opening section.
I may have talked about this here before & asked about it. Some piano players that I know, heard it & said that it was probably the noise of the pedal usage. It never sounded like a realistic sound to me - I always thought that it was some sort of anomaly, either in the recording or in the playback. Not so long ago I began to wonder if it was noise modulation in the playback system.
Anyway. this version of MQN is the first time that it is gone, I think - I haven't checked it on my main system just through headphones via my laptop.

Here's the intro for all to listen https://docs.google.com/file/d/0BzRUtXo ... sp=sharing
www.Ciunas.biz
For Digital Audio playback that delivers WHERE the performers are on stage but more importantly WHY they are there.
User avatar
DaveF
Posts: 2869
Joined: Sun Jan 17, 2010 10:38 pm

Re: MQN testing/experimentation thread

Post by DaveF »

jkeny wrote:
DaveF wrote:If equipment wasnt issue I'd be inclined to take the digital out just before the DAC and run it into a high speed ADC board connnected to a FPGA PCI-E card in a PC. Take several sample sets and use Matlab to do all the fancy analysis offline to your hearts content.
The ADCs would have to have a sufficient noise floor and high ENOB(effective number of bits) to make sure we have enough resolution.

Using scope to look at a stream in real time might be tricky to get a stable trigger.
But that will only be examining the digital logic signal, not the electrical waveform that is the physical form of this logic signal.
It's in this waveform that the noise is likely to be embedded & it's interaction with the USB receiver & downstream circuit may well be where there is some insight to be gained?
Sorry, typo in my earlier post. I meant of course taking the analog signal after the DAC otherwise no point in have an ADC! :-))
If the ADC has a high enough resolution we should be able to see the noise content of the signal also. If the PC is generating certain noise tones that are in any way dominant over the noise floor we should be able to see this.
"I may skip. I may even warp a little.... But I will never, ever crash. I am your friend for life. " -Vinyl.
Michell Gyrodec SE, Hana ML cart, Parasound JC3 Jr, Stax LR-700, Stax SRM-006ts Energiser, Quad Artera Play+ CDP
jkeny
Posts: 2387
Joined: Sun Jan 17, 2010 9:37 pm

Re: MQN testing/experimentation thread

Post by jkeny »

DaveF wrote: Sorry, typo in my earlier post. I meant of course taking the analog signal after the DAC otherwise no point in have an ADC! :-))
If the ADC has a high enough resolution we should be able to see the noise content of the signal also. If the PC is generating certain noise tones that are in any way dominant over the noise floor we should be able to see this.
Yea & I didn't read your post carefully enough to realise your mistake - sorry.
I know what you say makes sense but these are the sort of tests that Mithco & others have done already which have proven to be of no value in revealing differences (or as they have interpreted the results as very useful in "proving" that no differences exist)
I somehow feel that we are missing some psychoacoustic issue that is not being revealed in these tests. Some noise modulation that is below the noise floor is all I can think of. How would this be measured?

And, yes, we can hear below the noise floor
www.Ciunas.biz
For Digital Audio playback that delivers WHERE the performers are on stage but more importantly WHY they are there.
jkeny
Posts: 2387
Joined: Sun Jan 17, 2010 9:37 pm

Re: MQN testing/experimentation thread

Post by jkeny »

BTW, correction to my "Court & Spark" post above - it isn't as black & white as I first thought
The "noise modulation" that I hear on this track with other players (& other versions of MQN) is not completely eliminated in 2.66 SSE2 dqa dqa - it is only reduced. So not such a good candidate for testing, as I first thought.
www.Ciunas.biz
For Digital Audio playback that delivers WHERE the performers are on stage but more importantly WHY they are there.
jrling
Posts: 398
Joined: Tue Oct 08, 2013 7:54 pm
Location: London

Re: MQN testing/experimentation thread

Post by jrling »

jkeny wrote:
DaveF wrote: Sorry, typo in my earlier post. I meant of course taking the analog signal after the DAC otherwise no point in have an ADC! :-))
If the ADC has a high enough resolution we should be able to see the noise content of the signal also. If the PC is generating certain noise tones that are in any way dominant over the noise floor we should be able to see this.
Yea & I didn't read your post carefully enough to realise your mistake - sorry.
I know what you say makes sense but these are the sort of tests that Mithco & others have done already which have proven to be of no value in revealing differences (or as they have interpreted the results as very useful in "proving" that no differences exist)
I somehow feel that we are missing some psychoacoustic issue that is not being revealed in these tests. Some noise modulation that is below the noise floor is all I can think of. How would this be measured?

And, yes, we can hear below the noise floor
jkeny - the link I posted of Mitchco's test was comparing JRiver Win to JRiver Mac OS, so I am no that surprised they showed no difference. Both Intel-based CPUs but not nearly the same processors unfortunately.

But are you saying that someone has tried his type of test on different versions of MQn? If so, I don't think that I have seen the results?

Surely we need to eliminate the possibility that there may be demonstrable differences above the noise floor first?
jkeny
Posts: 2387
Joined: Sun Jan 17, 2010 9:37 pm

Re: MQN testing/experimentation thread

Post by jkeny »

jrling wrote: jkeny - the link I posted of Mitchco's test was comparing JRiver Win to JRiver Mac OS, so I am no that surprised they showed no difference. Both Intel-based CPUs but not nearly the same processors unfortunately.

But are you saying that someone has tried his type of test on different versions of MQn? If so, I don't think that I have seen the results?

Surely we need to eliminate the possibility that there may be demonstrable differences above the noise floor first?
From memory, Mitcchco did similar tests with Jplay Vs other players &/or there was another guy that also did similar tests, can't remember his sig. In all cases no differences found, I believe.
I did a null test between the output of MQN & the original file which found no difference down to -90dB but I need to repeat this test.
www.Ciunas.biz
For Digital Audio playback that delivers WHERE the performers are on stage but more importantly WHY they are there.
jkeny
Posts: 2387
Joined: Sun Jan 17, 2010 9:37 pm

Re: MQN testing/experimentation thread

Post by jkeny »

BTW, this thread is a very good read on hearing below the noise floor http://www.gearslutz.com/board/music-co ... floor.html

One quote that also stands out for me - from a norwegian mastering engineer Andreas Nordenstan - this post http://www.gearslutz.com/board/6049213-post19.html
We learn to recognize certain abstract aspects of sounds like we learn the more easily describable stuff like name of notes, instruments and effects like compression. Listening to the rest, whatever you want to call it, is a sort of listening that lends itself particularly well to intuitive work.

Think that's the primary reason many feel blind tests to be artificial compared to normal listening. It's enforcing the square world of words and truth tables onto the wordless language.

Have learned to somewhat successfully tune out/in when doing blind testing. The ABX results from listening consciously to this or that aspect of sound is consistently about 10-15 dB worse than when I do the "lateral listening".
www.Ciunas.biz
For Digital Audio playback that delivers WHERE the performers are on stage but more importantly WHY they are there.
jrling
Posts: 398
Joined: Tue Oct 08, 2013 7:54 pm
Location: London

Re: MQN testing/experimentation thread

Post by jrling »

I have noticed recently that versions of MQn compiled for SSE2 Instruction Sets have been getting some preference over identical versions compiled for SSE4 Instruction Sets. Even by those, unlike me, who have an SSE4 capable CPU.

My only uninformed guess is that SSE2 instructions may be less onerous for the CPU to process and therefore less CPU generated noise is the outcome?

Any views?

Jonathan
Maplin XM21X 12V float charging A123 26650 LiFePO4 battery/Maxwell Supercap PSU for Mitac PD10-BI J1900 Bay Trail, WTFPlay, Hiface Evo, Bow Technologies 1704 NOS DAC, StereoKnight TVC, Quad II monoblocks, ZU Audio Druid Mk4/Method Sub
jrling
Posts: 398
Joined: Tue Oct 08, 2013 7:54 pm
Location: London

Re: MQN testing/experimentation thread

Post by jrling »

I asked SBGK for any reason he thought 2.68 was sounding a step up in SQ and his reply was -

previously the loop would have a couple of prefetch instructions at the start which prefetched data for the next loop round, but this version fetches 8192 bytes into cache before entering the loop and then fetches the next 8192 after that buffer has been sent to the device, so that when the buffer is ready to be filled the data is already in cache, simple really. I was just experimenting, what I was trying to do was load 512 kb into cache so there was less disk access while rendering, but that wasn't possible, I could only load 8192 bytes, so this is really an accidental finding although it does make sense to do it this way.

Don't know if that gives anyone further intelligence into why different MQn versions' SQ differ from each other?
Maplin XM21X 12V float charging A123 26650 LiFePO4 battery/Maxwell Supercap PSU for Mitac PD10-BI J1900 Bay Trail, WTFPlay, Hiface Evo, Bow Technologies 1704 NOS DAC, StereoKnight TVC, Quad II monoblocks, ZU Audio Druid Mk4/Method Sub
jesuscheung
Posts: 2491
Joined: Mon Oct 07, 2013 11:09 pm

Re: MQN testing/experimentation thread

Post by jesuscheung »

sorry guys. i don't understand the point of measurement.... if a complete and sound measurement exists, then every DAC on earth would have been ranked mathematically.

the fun of hifi is listening. why reason it with measurements? coz our ears aren't reliable? if so, why bother with hifi...

sorry to spoil the fun. i guess there is a point. programming compiler have well-defined test cases. yet player compiled by newer compiler seems to produce better SQ. yet, compiler's test cases weren't goal for better SQ, but for better performance and stability.

also, when you start to measure 'cache misses', i feel that you are intercepting, hence damages SQ, hence the measurement is not the true MQn audio output. it is like a firewall intercepting every packet going in and out of your computer.
Post Reply