[Woem]

You really need to get your ears and/or head checked. The only thing that ReplayGain changes is the digital volume. It's exactly like using the Volume Control DSP. There may be some change in the way the quantization noise works, but that'd be at the bit-depth you're outputting at, which is beyond the noise floor of your equipment if you're using 24 or 32-bit output.

ReplayGain is exactly analogous to turning your volume knob down because you find one track's louder than the next. It's just an automated way of doing that.

Note to anyone else concerned with the problem PatrickHat is having: it exists only in his head. It is linear attenuation. It cannot affect the tonality of sound, nor alter its character, colour, or any other subjective quality of sound other than volume.

 

[Publius]

Thank you, Woem, for your reply. If all of us could be as timely, informative, and persuasive!

 

[Patrickhat2001]

Not this thread again, my Head-Fi fall from credibility.

Anyway I got to the source of the problem a few weeks ago--it was my ASIO buffer. I had it disabled, and this caused any extra processing to produce jitter and hence degrade the signal (or, at least, sound different). At least that's my reasoning and it's what's worked for me--since I enabled the buffer I haven't noticed a difference between resampler being on or not, nor do I hear differences between lossless files and waves anymore (my credibility went where again?

). I'm willing to bet I won't hear the differences between ASIO and Kernal Streaming anymore either (not a suprise because Kernal Streaming is said to be the same thing as ASIO except with a buffer built in) but I haven't gotten around to testing this yet. I don't know what the official word on the street is about the effects of the ASIO buffer being enabled/disabled (supposedly there are multiple buffers besides the one found in ASIO) but with it disabled any extra processing (or even slight differences in how files were decoded such as with lossless vs waves) seemed to produce a difference in sound. Now that buffer is enabled, bam, no differences and it's been that way for a few weeks now.

Sorry I never did get around to that blind testing--I couldn't talk any of my friends into it.

P.S. I stated earlier that I thought I could tell small differences between having resampler enabled/disabled with waveout in a quick twenty minute listening test where I was also testing a lot of other conditions. I am willing to concede that those differences were likely in my head (I thought they were very slight anyway).

 

[Stephonovich]

Quote:

Originally Posted by Publius Thank you, Woem, for your reply. If all of us could be as timely, informative, and persuasive!

Ah... I see he's returned. The author of such thoughtful, mind-blowing posts such as...

Quote:

Originally Posted by Woem You, sir, are an idiot. Shut up and re-evaluate your alleged ear for music. You, on the other hand, I will call technically incompetent... That said, if you're going the external soundcard route, stay away from MP3. WMA reproduces the tonal purity and balance of music much better than MP3. Microsoft spent around $27 million properly tuning WMA, while most MP3 encoders are tuned in some geek's spare time.

Ah... good stuff. I especially like the last one. I'd love to see that posted over at HA. Wonder what Dibrom would have to say, or even better, Garf?

 

[Wodgy]

Quote:

Originally Posted by Woem You really need to get your ears and/or head checked. The only thing that ReplayGain changes is the digital volume. It's exactly like using the Volume Control DSP.

Don't you just love it when someone who has not completely thought through the underlying algorithm goes off and insults someone?

You're wrong, Woem. While ReplayGain is implemented in a simple multiplicative fashion (the actual details are here: http://replaygain.hydrogenaudio.org/player_scale.html ), in any listening session involving more than one song it does have perceptual effects because it alters the baseline loudness from one song to another. As discussed here:
http://replaygain.hydrogenaudio.org/..._clipping.html
"Replay Gain will make loud dynamically compressed tracks quieter, and quiet dynamically uncompressed tracks louder. The average levels will then be similar, but the quiet tracks will actually have louder peaks."

Then, of course, there is the issue of ReplayGain-induced clipping. Foobar does not have a graceful solution to this, other than the user realizing there is clipping and choosing to lower the preamp slider. In my experience, most users choose to invoke the Advanced Limiter to solve this problem, rather than do the obvious, simple solution which is to lower the preamp slider. The Advanced Limiter will have audible effects.

Quote:

Note to anyone else concerned with the problem PatrickHat is having: it exists only in his head. It is linear attenuation. It cannot affect the tonality of sound, nor alter its character, colour, or any other subjective quality of sound other than volume.

Again, tonality is perceived in relative terms, just as volume is. ReplayGain alters the amount of compression per song relative to other songs, and thus does tend to affect perceived tonality, even when no clipping is occurring.

 

[Woem]

Quote:

Originally Posted by Wodgy Again, tonality is perceived in relative terms, just as volume is. ReplayGain alters the amount of compression per song relative to other songs, and thus does tend to affect perceived tonality, even when no clipping is occurring.

Ehm, no. You're wrong. ReplayGain does no compression of any sort. Yes, it's multiplicative. Multiply by a constant, add/subtract on a logarithmic scale, apply a 1-element FIR filter... They're the same thing. Go higher math.

foobar2000's implementation is easy to understand, and completely compliant with the spec. It's a simple volume-changer. Someone complained about how fb2k doesn't have an elegant method of handling clipping induced by replaygain. Again, wrong. It's certainly got the most elegant method of any player out there. On top of supporting universal ReplayGain, it also allows >1.0FS values for samples, which allows for graceful degradation using the hard limiter and advanced limiters.

It does not change the harmonics of anything, does not affect any frequencies, and so on.

In other words: you lose. Learn some simple audio signal processing and try again next time.

 

[Publius]

Quote:

Originally Posted by Woem Ehm, no. You're wrong. ReplayGain does no compression of any sort. Yes, it's multiplicative. Multiply by a constant, add/subtract on a logarithmic scale, apply a 1-element FIR filter... They're the same thing. Go higher math.

Did Wodgy's explanation go entirely over your head?

He was talking about this. It is basic psychoacoustics.

I have to disagree with Wodgy's interpretation, because the intensity of the music is really only set by the listener and not the music itself for most situations, so the perceived pitch will always be off. But that's no excuse to just troll (something you're apparantly rather apt to do).