[Kuja]

Quote:

Originally Posted by Riordan publius, kuja: calling someone a troll for giving his experience and asking questions is rather rude. citing other sites' rules that (for good reason!) have no relevance on head-fi is not helping either. this is NOT a scientific community, nobody makes scientific claims, nobody needs to prove or disprove anything.

I didn't call him a troll. I simply thought his post was indicative of the level of wishful thinking that goes on around here. Your last sentence pretty much sums up the worst type of audiophile that exists - one who enjoys a hobby based around paying exorbitant fees for placebo-effects, claims that cannot be substantiated in any way, and yet is proudly defensive of the lack of scientific testing for the simple reason that the monumental edifice of self-delusion would come crashing down in the face of ABX tests that reveal nothing beyond random chance.

Hell, just for a little fun, why doesn't someone post this thread at Hydrogenaudio and see if we can't get the RG devs to show that the claims made here are for all intents and purposes purely in the realm of 'fairies at the bottom of the garden'.

 

[JeffL]

Quote:

Originally Posted by Kuja I didn't call him a troll. I simply thought his post was indicative of the level of wishful thinking that goes on around here. Your last sentence pretty much sums up the worst type of audiophile that exists - one who enjoys a hobby based around paying exorbitant fees for placebo-effects, claims that cannot be substantiated in any way, and yet is proudly defensive of the lack of scientific testing for the simple reason that the monumental edifice of self-delusion would come crashing down in the face of ABX tests that reveal nothing beyond random chance. Hell, just for a little fun, why doesn't someone post this thread at Hydrogenaudio and see if we can't get the RG devs to show that the claims made here are for all intents and purposes purely in the realm of 'fairies at the bottom of the garden'.

Statistics and specifications are one thing, Kuja, reality is a totally different matter.

Any type of signal processing, on the level of digital or analog, will affect the sound in some way.

It is possible that the problem is ReplayGain, but that is highly unlikely, however not out of the question! It is possible that ReplayGain is causing a jitter problem, or even causing a problem with the headphones/amp.

Just because you can prove RG doesn't affect the sound quality, doesn't mean that it won't, whether a direct effect or indirect.

Quote:

Your last sentence pretty much sums up the worst type of audiophile that exists

Quite the contrary. How is it that a $25 Chaintech is preferred by some members over a EMU 1212m/404? Yes, you could easily do an ABX test, if you are so inclined. But as I said, science is and never will be the end -all be all!

I'd say you are bordering on trolling yourself, consdering your grandiose comments here, and bold assumptions, without fact or "science" to back them up

 

[Riordan]

kuja,
no, you didn't call him a troll - publius did, or nearly did. i apologize for summing up two replies into one.

yes, a lot of "wishful thinking" goes on here. as i said, this is no scientific audio community. there are others for that, hydrogenaudio being the most reliable that i know of, but also the most unfriendly.

my last sentence simply sums up the vast majority of internet (or any human) communication and has nothing to do with the "spending for the sake of spending" or audiophile mumbo-jumbo that you seem to read into it.

there is a place for scientific audio research and controlled abx testing - but to scold someone (or a forum community) simply for not doing it is, again, rather rude. on head-fi SUBJECTIVE aspects of audio enjoyment are discussed. you were absolutely right in your post - if head-fi adopted hydrogenaudio's rules, precious little would be discussed at all, because you cannot objectify subjective impressions.

every once in a while plain mistakes are made, or things are claimed as fact, not opinion, that don't hold up to scrutiny.

this is where knowledgeable folks come in and correct the mistakes - thanks for that. but please understand that this is NOT hydrogenaudio and most of us are very glad about that. around here, civility and respect is mandatory - proof is not

(and just to make sure you don't misunderstand, i personally got nothing against the proof part - it's the loss of civil discussion that gets to me...)

 

[Publius]

First of all, I apologize (again) for throwing around words like trolling without really having good cause to. Second I apologize for being too hard on Patrick for requesting more evidence. He's clearly got he best of intentions here.

Still, I think this is worth resolving one way or the other, and I regret getting this thread off course. Either there really is something wrong with RG, or there isn't, and any confusion one way or the other is going to misinform a lot of people. Just because we aren't supposed to be the infallible source of information on this site doesn't mean people are still going to treat it as such.

Quote:

Originally Posted by JeffL Any type of signal processing, on the level of digital or analog, will affect the sound in some way.

In the digital domain this is trivially wrong - in the floating-point domain, amplify a signal by an amount, then attenuate it by the negative amount, and you get the original signal within the errors that the precision allow. Even in single-precision floating point with a 24-bit mantissa, the error is 20dB under the noise floor of any analog I/O equipment. And this is, more or less, exactly what ReplayGain is doing, except it's doing it with at least 52 or so bits of precision, not 24....

In the analog domain...
Quote:

It is possible that the problem is ReplayGain, but that is highly unlikely, however not out of the question! It is possible that ReplayGain is causing a jitter problem, or even causing a problem with the headphones/amp. Just because you can prove RG doesn't affect the sound quality, doesn't mean that it won't, whether a direct effect or indirect.

There actually is a vector for how this could happen. In a quiet enough environment, probably even with high-end cards, you ought to be able to hear foobar's processing thread wake up every millisecond or so. Try it! Get your environment as quiet as possible, crank up the volume as high as it will possibly go, then let foobar play some low-level silence, something really, really quiet, but not completely silent, preferrably in a format that uses a lot of CPU time to decode. I've actually heard this maybe once, and it was mobo audio.

But just because a problem may exist doesn't necessarily mean that a problem does exist. The plausibility of any hypothesis about RG affecting power rails or jitter or whatnot, as well as the audibility of such, always needs to be weighed against the null hypothesis - ie, placebo. And the audibility of power line noise while full volume music is playing is questionable enough, much less second-order effects such as intermodulation.

Which brings up the other troublesome point I have with this thread, which is really what worked me up in the first place. Say that I'm forced to personally conclude that all Patrick noticed was placebo, and his initial observations are invalid. In the absence of any credible vector for inducing this effect, or objective proof, this is a reasonable conclusion to make. What I am now to make of any statements he makes in the future about tonal character and detail? If he was mistaken before, why would he not be mistaken again? This sort of thing reduces my respect for his opinions, which is exactly what I'm trying to avoid.

Or RG itself could just have a bug in it. But if it does have a bug, it should be relatively easy to identify by tweaking configurations and preprocessing the data, which is why I've been asking Patrick for more information on this, and I'm surprised that anybody else is not as curious about it as I am. This isn't some magic algorithm processing your sound, it is elementary school multiplication. This ought to be easy enough to troubleshoot, even in a subjective context... but it gets easier with objective testing.

 

[Patrickhat2001]

The ol' brain is fried from school and I find myself not really being able to think so I hope I can sort out my thoughts to say what I want to say effectively.

First of all I'm sorry that it has taken me so long to respond, I just haven't had time as of late. I also haven't had enough time to listen to my rig lately so it looks like the ABX comparisions will have to wait until this weekend.

I agree with all of the criticisms made thus far. I, too, want to keep the quest for high quality audio as much in the scientific realm as possible and and I consider myself very much a sceptic concerning just about everythinig related to this hobby. For example I don't believe in burn in (well, unless you meen in a purely psychological sense), high priced interconnects or even in spending all that much money on any type of equipment since I think it's only a matter of time until someone else figures out why something sounds good and starts to make the same or a similar product for less (the $25 Chaintech is a good example of this). I raise an eyebrow, too, whenever someone mentions something like cryo freezing interconnects or cooking cables. I hate bulls#%t and don't tolerate it.

Which brings me to my point--many people no doubt saw what I was saying about replaygain as bull. I understand that reasoning. Afterall, as the saying goes, digital is digital is digital--as long as the digital signal itself isn't changed before it's converted into analogue there soundn't be a change in the sound. Personally, I have no doubt that, even with replay gain enabled, the playback of music is still bit perfect (ie nothing changed or missing). What I think is happening, though, is replaygain is effecting the timing of the playback (aka adding jitter). Such a thing is not a new idea--it's a common recomendation around here that users should disable all the DSPs within Foobar unless there actually something that they use (volume setting, crossfeed, etc). Why is this recommendation made? Because the more settings that Foobar (or any other player) has to cycle through while playing an audio file the more liable it is that the timing of the playback of the file will be changed slightly. This is exactly what I think is happening when replaygain is enabled. All the sounds of a song are still there, nothing is missing, but things sound different. The sound isn't as transparent nor does it flow as effortlessly as it should. To give an example of what I mean all use a favorite track of mine Tenacious D - "Inward Singing."

The song begins with Kyle Gass snoring away for about twenty seconds. With replaygain disabled the sounds of snoring flow together in a very natural way--effortlessly, much like real life, and the sound is very transparent. But enable replaygain and each breath and snort is more distinct from the other but in an unnatural way--it doesn't sound right and you can recognize it when you hear it (sorry for the crummy explanation, I told you I couldn't think too well right now

) and the sound is overly dense; lacking transparency. This lack of transparency in the sound and lack of natural flow are what I meant when I said things sounded digitized in my post above. The stuffyness of the sound was also what I was referring to--incorrectly--in my first post when I mentioned tonal character. My bad. Sorry about that. Although the sound of instuments is slightly changed with replaygain is enabled it's really because transparency is lacking--characteristics like warmth or neutrality are not effected. I don't know why I even mentioned tonal character in my above post, it was stupid to do so.

The differences that I hear may not be a big deal to some people and frankly I wouldn't be suprised if others didn't notice them. However, I certainly do, and being an audiophie they bug me to death.

Something that I also thought of which could be causing the change in sound could be my computer itself. I'm clunking along with an 800Mhz Celeron with 128 MB of ram running Windows XP. This isn't exactly the most state of the art set up and it shows--when I'm doing many things on my computer the card will sometimes make really annoying buzzing sounds when playing music and sometimes will stop playing music entirely. Perhaps, because I am using a clunker enabling replaygain blogs it down ever so slightly and effects the speed of playback whereas this wouldn't occur at all on faster computers.

 

[Publius]

Hey - we are all reasonable men here. We don't need to give assurances as if we were lawyers.

Quote:

What I think is happening, though, is replaygain is effecting the timing of the playback (aka adding jitter). Such a thing is not a new idea--it's a common recomendation around here that users should disable all the DSPs within Foobar unless there actually something that they use (volume setting, crossfeed, etc). Why is this recommendation made? Because the more settings that Foobar (or any other player) has to cycle through while playing an audio file the more liable it is that the timing of the playback of the file will be changed slightly.

They also suck up CPU time and thus make your system slower in general, if imperceptably. This is generally a better reason for a lot of people.

One thing I'm not sure you're aware of is that on any sort of video card - even old ISA ones - the CPU and/or OS is not responsible for timing the audio - the sound card is. There are a bunch of ways the audio actually gets down there - basically, either the OS maintains a buffer and the card DMAs (requests) it occasionally, or the OS interrupts occasionally to send said buffer, or maybe the OS even PIOs the buffer (sends it across slower). But however it gets sent across, it gets sent as a buffer, which gets stored on the card, and emptied into a DAC at a fixed rate. Nothing in this system has any regard for how the CPU can affect the sample rate of the card, because the only way that happens is if the OS programs it too. If the card misses a sample, that doesn't result in any sort of slight sound changes, it results in a click, and unless the computer is always bogged down, it is definitely a bug.
And of course, if the sound is merely "delayed", that doesn't affect the sound, because each sample would get delayed by the same amount.

Finally, if it is causing some sort of analog effect on the sound card, like through the power rails or what not, it is going to be very, very, very light. But it would be quantifiable, perhaps by recording the output to a separate computer with an equally high quality card.

Again, if CPU usage or quantization or anything of that sort really is the problem, it would readily be comparable by discwriting the RG'd file to wav, then amplifying it back to the original volume level, and comparing source and output.

 

[Kurt]

Do you have onboard LAN or RAID? If so, try with disabling it.

 

[ReDVsion]

and here I thought my music sounded BETTER with rpg... but hey, to each his own!

 

[Mr.Radar]

Argh! I probably can't really tell a difference but now the placebo effect has kicked in and I can't listen with RPG enabled anymore, however I hate having to adjust my volume control all the time, especailly since the gain on my CHA47 is set too high for my headphones.

 

[Riordan]

just an idea out of the blue: i've read recommendations to disable any plugin in foobar for audigy cards because having them enabled leads to the card resampling - could something like this be the case with the emu, too? i doubt it, because patrickhat2001's impression is, if not a first, a minority one... but maybe replaygain is not the direct but an indirect culprit? is there anything else that could affect the sound based on turning replaygain on/off?

mind you, "very little" difference for one could well be "night and day" for another, and aural placebo IS a powerful force to be reckoned with

 

[gorman]

Publius, as you seem to be knowledgeable enough... I ripped my CD collection to FLAC, replaygained it. I use FLACs for listening on my computer. But Vorbis Q6 for my Karma.

Since Karma (and no other portable player that I know of) does not support Replaygain, I converted from FLAC to Vorbis with OggDropXP, scaling them using the option to use the replaygain tags.

My question is: by doing this am I actually losing parts of the sound information and, thus, sound quality?

 

[Publius]

Me? Knowledgable?

We're discussing two different potential effects here: quantization noise in the digital output itself, and an electrical effect on the real-time sound card output. The former is definitely happening, and it does effect your oggs, but it is more or less eliminated if you use dithing. The real debate, IMHO, is over the second, since dithering does not appear to be helping this.

Sampling rates are explicitly converted in foobar's DSP chain, and there is no chance that replaygain would be messing with it. My RME does show sampling rate changes with PPHS/SSRC but not with any other component that I know of.

 

[tschanrm]

woops, didn't really mean to reply to this old post, it was late and I didn;t realize the topic was over two months old, oh well....


Quote:

Since Karma (and no other portable player that I know of) does not support Replaygain, I converted from FLAC to Vorbis with OggDropXP, scaling them using the option to use the replaygain tags. My question is: by doing this am I actually losing parts of the sound information and, thus, sound quality?

Well, if you mean converting flac (lossless) to Vorbis (lossy), then I'd say you are definitely losing sound quality, but if you have a high vorbis bitrate you'd be hardpressed to ABX most material.

If by you mean replaygain is causing loss of sound information, I'd say definetly no, at least when it comes to ripping the file and adding a replaygain tag. Processing replaygain for a song I have no idea, just assumed you didn't lose information, but thats what this thread is about right? Replaygain works just like id3 tags when you are ripping a song, they are both metadata, or they are added on to the end of your music file. When a player plays a music file but doesn't support the metadata, it simply ignores the metadata completly and plays the file. This can happen with old mp3 players that are not id3v2.0 compatible. So to anwser you question (if this was your question), replay gain does no information loss when you scan a file with replaygain. All replay gain does is scan over the song, find the right db adjustment for it to be at the standard gain, and store that db adjustment as metadata (album and track gain info). I'm not real sure how replaygain works when you are actually playing a song though, sorry. Hope this info helps, and that I've answered somebody's question

 

[MikoLayer]

Quote:

Originally Posted by Kurt Do you have onboard LAN or RAID? If so, try with disabling it.

just curious, how does that affect the sound? IRQ problems maybe? you made me wonder if i should disable my nforce2 onboard NIC

 

[Stephonovich]

I've always used RG, although my main interest was it's implentation in MP3Gain for my various MP3s I've collected. Come to think of it, I have no idea why I have it enabled in Foobar. None of my albums clip, and I don't mind having to adjust volume between them. It's just adjusting between each song that got to me, hence, MP3Gain.

As for SQ loss, um, no... I can't see that happening. To each his own, and I'm not discounting your experience, but I just can't see how it'd be feasible. I'm not putting anything past computers/Windows, however...