FORUMS
Remove All Ads from XDA

[MOD][TWEAK][03/05] Nexus S: Fidelity 2.0 - Ultimate Low latency audio playback

742 posts
Thanks Meter: 1,304
 
By Windows X, Senior Member on 27th April 2012, 05:47 AM
Post Reply Email Thread
4th May 2012, 07:11 PM |#111  
polobunny's Avatar
Senior Member
Flag Montreal
Thanks Meter: 2,495
 
More
Quote:
Originally Posted by Windows X

kong, I already answered your questions and wrapped it into one post. In my defense from many unwanted guests trying to make it sounds absurd to try, I have to make my words clear that I'm making low latency audio patch not better sound for voodoophile. I apologize if that sounds offensive to you and I already said I'm more appreciated to help in explaining about low latency thing.

As for bedalus' remarks, low latency is one of the most promising measurements worth chasing in world of pro audio and computer audiophile. I don't see why it's wrong to use scientific pro audio terms that audiophiles can make some sense with. And sonic degradation suffered from lower latency is real thing because random jitter added in latency will be wider. But you have to know how latency jitter can cause problems in quantization. It's pretty deep and easiest way is just try it and decide if you want to keep it or not. I've been in this circle numerous times already in computer audiophile.

I hope to hear your test report soon as I'm also looking forward to see how it'll work in your case too. Personally, I don't see anything wrong in here trying to have lower latency for audio playback regardless of getting better sound quality or not.

For what it's worth, I don't think anyone here has a problem with reducing the latency, rather with the claims that were made of possible audio improvements while reducing the latency.

I'm interested in what Bedalus can find though. While I am in no way a certified audio engineer, I have done more than a few installations of professional hardware (MOTU & M-Audio mostly) and worked with professional software (AVID) and fully understand the concept of reducing the latency in lag sensitive cases.
I do not however ever remember encountering an instance where the latency proved to cause deterioration of the sound quality using DACs.

Either way, as I said twice before, to each their own. If you can reduce the latency while keeping the overheard to a practical level (not having to use the performance governor) then I see nothing wrong with that. At worst, it'll improve the apparent responsiveness of the device.
The Following User Says Thank You to polobunny For This Useful Post: [ View ] Gift polobunny Ad-Free
 
 
4th May 2012, 08:50 PM |#112  
Senior Member
Flag Irvine
Thanks Meter: 39
 
More
One interesting thing regarding the lower latency which could be really beneficial is audio production apps such as caustic. The main problem with android in the past is that there is quite a noticeable delay from when a software button is pressed to when the sound is heard. It would be awesome if this mod could fix that and make it more like ios where there is almost no discernible gap. I'll report back when I have a chance to try it out. I haven't read through the whole thread so sorry if somebody as already mentioned this.

---------- Post added at 12:50 PM ---------- Previous post was at 12:17 PM ----------

So I've tried Caustic with this mod and am very happy to report that there is no discernible time between the button press and the sound through the headphones which is amazing. I've been waiting for something like this to come around for a long time. Now this is on par with any ios device in regards to that. Really fantastic work =)
4th May 2012, 09:51 PM |#113  
polobunny's Avatar
Senior Member
Flag Montreal
Thanks Meter: 2,495
 
More
Quote:
Originally Posted by tycruickshank

One interesting thing regarding the lower latency which could be really beneficial is audio production apps such as caustic. The main problem with android in the past is that there is quite a noticeable delay from when a software button is pressed to when the sound is heard. It would be awesome if this mod could fix that and make it more like ios where there is almost no discernible gap. I'll report back when I have a chance to try it out. I haven't read through the whole thread so sorry if somebody as already mentioned this.

---------- Post added at 12:50 PM ---------- Previous post was at 12:17 PM ----------

So I've tried Caustic with this mod and am very happy to report that there is no discernible time between the button press and the sound through the headphones which is amazing. I've been waiting for something like this to come around for a long time. Now this is on par with any ios device in regards to that. Really fantastic work =)

I was thinking that but did not know there was any audio production app worth mentionning on mobile. Cannot lie, I don't use my phone for anything out of the ordinary. Games, web, text and calling.
bedalus
5th May 2012, 12:33 AM |#114  
Guest
Thanks Meter: 0
 
More
Here are the data from my test:

control
-3.43 signal
-78.5 noise
=75.07 difference

extreme insane
-5.37 signal
-75 noise
=69.63 difference

extreme direct
-5.37 signal
-80 noise
=74.63 difference

extreme linear
-5.37 signal
-76.5 noise
=71.13 difference

(all measurements are RMS power in dBFS.) As we can see, the mod provides no improvement in the signal to noise ratio. The control test (no mod) was actually the best (but not significant beyond any reasonable error margins).

Clearly this shows that there is no noise reduction provided by this mod. In fact, despite setting the performance governor, I experienced audio glitches with each of the three mods (but not in the control test).

Methodology

I produced a 44.1kHz 24 bit stereo lossless WMA file of a 1kHz sine wave at -6dBFS, with one channel inverted, with appropriate dithering (provided by Waves IDR, type 1, normal noise shaping). I then recorded the output into pro tools, then routed both L and R channels into a mono aux channel, making fine gain adjustments to completely cancel out the sine wave, leaving only noise. (RMS metering was provided Elemental Audio's IXL RTAS plugin.)

So, in summary (again...) longer latencies do not introduce jitter.
The Following User Says Thank You to For This Useful Post: [ View ] Gift Ad-Free
5th May 2012, 01:10 AM |#115  
OP Senior Member
Thanks Meter: 1,304
 
More
What about plus_general and plus_music? For what it's really worth, plus_general is stable for common apps use and plus_music is far more stable than extreme ones. I never once said jitter will introduce more noise. Latency jitter is something you need to measure for jitter if you want to see that caused it or not and CMP already did that.

The reason high latency is bad for audio output quality is burst of electrical activity. The larger buffer is, the longer burst runs giving more interference to soundcard (and also jitter). Plus, if you use FLL (anti-jitter in Voodoo), lower latency means more frequent of signal feeding to loop making FLL more efficiently used.

Anyway, your researched has something interesting for signal value. Your signal value after patching changed 56.57% while noise remains about the same. Why did that happen?
bedalus
5th May 2012, 06:43 AM |#116  
Guest
Thanks Meter: 0
 
More
Quote:
Originally Posted by Windows X

What about plus_general and plus_music? For what it's really worth, plus_general is stable for common apps use and plus_music is far more stable than extreme ones. I never once said jitter will introduce more noise. Latency jitter is something you need to measure for jitter if you want to see that caused it or not and CMP already did that.

Jitter causes noise. Are we talking about jitter as the bitstream gets reclocked or not?

Quote:
Originally Posted by Windows X

The reason high latency is bad for audio output quality is burst of electrical activity. The larger buffer is, the longer burst runs giving more interference to soundcard (and also jitter).

Interference is a type of noise, and measurable. Again, I didn't measure any difference. You say jitter, so I assume you mean this thing you call "latency-jitter" (it would be easier if you stopped mixed up real terms with fantasy ones), which doesn't exist.

Quote:
Originally Posted by Windows X

Plus, if you use FLL (anti-jitter in Voodoo), lower latency means more frequent of signal feeding to loop making FLL more efficiently used.

Everybody should have anti-jitter enabled unless the system clock is more stable than the Wolfson clock, which would be hard to believe. Anyway, it doesn't matter how big or small your buffer is, bits go to the FLL one by one.

Quote:
Originally Posted by Windows X

Anyway, your researched has something interesting for signal value. Your signal value after patching changed 56.57% while noise remains about the same. Why did that happen?

After flashing your mod my volume set itself to a higher value. Anyone who has reported better sound quality from this mod (which as far as I can tell is only psychologically possible, rather than technically possible) has simply been fooled by hearing a louder sound.
The Following 2 Users Say Thank You to For This Useful Post: [ View ] Gift Ad-Free
5th May 2012, 07:02 AM |#117  
OP Senior Member
Thanks Meter: 1,304
 
More
What I'm trying to point out is latency and noise are difference things to measure. It makes no sense from you saying you want to measure about latency jitter and give me signal and noise report. Where's latency measurements? Read http://www.cicsmemoryplayer.com/inde...PeriodicJitter in Jitter research part carefully and do some experiments like his if possible.

Judging from how you talk about FLL, you don't really know how it works at all do you? The more hz of buffer streaming feeding to loop, the more accurate loop can lock. It's also stated in http://www.cicsmemoryplayer.com/inde...ay.ASIOLatency but as PLL which use similar concepts of locked loop.

For 'latency jitter' definition, go ask Emm labs, dCS, RME, Metric Halo and Weiss engineers whether it's fantasy word or not. There's more than what you think it has to be done to make stuff like tens of thousand bucks source worth it.

Again, there's clear evidence that lower latency can give benefit to improve locked loop features like FLL/PLL and minimize range of periodic jitter in given links. But using the word 'fooled' was a bit hypocrite as if you don't really care to learn anything new from it and offended all the people who said the opposite to you. I don't like continue talking in this manner.
bedalus
5th May 2012, 07:22 AM |#118  
Guest
Thanks Meter: 0
 
More
Would you explain what a FLL does please?

EDIT:

http://www.cicsmemoryplayer.com/inde...PeriodicJitter

Good article. You want me to do an spectrum analysis? I can do that. But this article doesn't mention 'latency-jitter'.

http://www.cicsmemoryplayer.com/inde...ay.ASIOLatency

Yeah, I remember this link from the last time you posted it. You didn't respond to any of my numerous criticisms of it. Here they are: http://forum.xda-developers.com/show...7&postcount=21
5th May 2012, 07:52 AM |#119  
OP Senior Member
Thanks Meter: 1,304
 
More
Your criticism doesn't sound like you know what you're talking with so I left it unanswered. The relation between those two links is ASIO Latency talking about periodic jitter and that link explains how periodic jitter can ruin up digital domain.

for FLL, I suggest you doing some google search and coursework first and I'll help explaining anything you stumble upon if I can.
5th May 2012, 08:00 AM |#120  
wildestpixel's Avatar
Senior Member
Flag London
Thanks Meter: 574
 
More
Quote:
Originally Posted by Windows X

Your criticism doesn't sound like you know what you're talking with so I left it unanswered. The relation between those two links is ASIO Latency talking about periodic jitter and that link explains how periodic jitter can ruin up digital domain.

for FLL, I suggest you doing some google search and coursework first and I'll help explaining anything you stumble upon if I can.

I'm with Bedalus here particulalrly as a statistician, R & D fellow and coder.

Editing a source file to decrease a value is something I do on a basic level with SystemUI.apk.

Supercurio has got the whole out of this device, don't effing patronise him, Bedalus or in turn myself thankyou.
The Following 4 Users Say Thank You to wildestpixel For This Useful Post: [ View ] Gift wildestpixel Ad-Free
5th May 2012, 08:43 AM |#121  
kong's Avatar
Senior Member
Thanks Meter: 958
 
More
Quote:
Originally Posted by bedalus

After flashing your mod my volume set itself to a higher value. Anyone who has reported better sound quality from this mod (which as far as I can tell is only psychologically possible, rather than technically possible) has simply been fooled by hearing a louder sound.

That's what my low-fidelity ears observed as well.

Quote:
Originally Posted by Windows X

Your criticism doesn't sound like you know what you're talking with so I left it unanswered. The relation between those two links is ASIO Latency talking about periodic jitter and that link explains how periodic jitter can ruin up digital domain.

for FLL, I suggest you doing some google search and coursework first and I'll help explaining anything you stumble upon if I can.

What happened to this?
Quote:
Originally Posted by Windows X

... But if you want to learn about low latency things, I'm more appreciated to help explaining about it. ...


Why am I quoting myself here?
Quote:
Originally Posted by kong

... But it seems my questions never got a direct answer nor an "average Joe" understandable explanation. ...

Déjà vu
The Following 2 Users Say Thank You to kong For This Useful Post: [ View ] Gift kong Ad-Free
Post Reply Subscribe to Thread

Tags
audiophile, head-fi, libaudio, mod, nexus s

Guest Quick Reply (no urls or BBcode)
Message:
Previous Thread Next Thread
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes