Resampling Question

[chia-pet]

What are the merits of resampling at 44100? Does anyone have an idea of whether this is better, worse, or whether there's no difference at all when compared to not resampling at all?

On a similar note, does anyone know why a "flat" setting is provided on my ipod? I would think turning off the EQ equates to a "flat" setting, or am I wrong? Seems redundant to put that there.

Thanks head-fiers.

 

[BrokenEnglish]

http://www5.head-fi.org/forums/showthread.php?t=95415

 

[chia-pet]

Correct me if I'm wrong, but doesn't that thread focus on resampling to something higher? Doesn't your soundcard (for most people) automatically resample to 44100 without the assistance of software?

So my question was, what are the merits of resampling to 44100?

 

[llmobll]

Nothing. If you send audio recorded at 44100 then resampling would just use extra CPU cycles for nothing. Its like trying to make your water more watery. Now resampling to a higher rate, well thats a whole other discussion.

 

[chia-pet]

Quote:

Originally Posted by http://ubernet.org/modules.php?name=Forums&file=viewtopic&t=871 -set target sample rate to 48000Hz (NOT to 192000 or anything like that. The reasoning is that all 10k2 based cards (as all creative cards are...audigy, audigy 2, etc, as well as all AC'97 cards, wich include nearly everything onboard, etc) resample EVERYTHING to 48000Hz in the hardware. So it wouldn't matter what samplerate you set it to, it's all being turned to 48000Hz. BUT, a problem arises....the hardware resampler does a crappy job. The solution is to set a GOOD software resampler (like the SSRC algorithm....included in foobar) to resample to 48000Hz, therefore bypassing the hardware resampler.

I found this on a foobar2000 setup guide. Any interpretations?

 

[llmobll]

Quote:

Originally Posted by chia-pet I found this on a foobar2000 setup guide. Any interpretations?

yeah use search, this topic has been beaten to death. The creative audigy line cards resample to 44100 because of their hardware and algorithms. not every card does this.

 

[Clown12]

It's true. I've gone right through my setup and it has these stages:
1) Foobar/winamp etc
2) Windows Audio, kmixer
3) WDM audio driver (written by me

)
4) FM801 PCI controller chip on sound card
5) WM9704Q DAC chip on card
6) STAC9721 DAC for surround speakers

The WM9704 and FM801 chips both do 48000Hz and 44100Hz. The STAC9721 doesn't.
This means the FM801 is forced to use 48000Hz all the time (regardless of me changing every single setting/disconnecting several traces etc, it just won't enable 44100Hz)
The FM801 SRC (hardware conversion) is more like a noise generator (seriously), so in my driver I only report 48000Hz and force kmixer to do the SRC.

The AC97 specification defines 48000Hz as mandatory, and 44100Hz as optional, so while many DACs might support 44100Hz not many controllers+drivers will support it or bother enabling it, since kmixer handles this pretty well for you anyway. Remember it's all commodity hardware.

 

[Stephonovich]

Dunno why no one's answered this for you yet...

Resampling to 44.1KHz is never a good idea, unless you have really low source material and want to bump it up for some reason. The article snippet you posted referred to resampling to 48KHz will benefit you if you have a card that resamples to 48KHz internally, i.e. every card Creative's come out with, with the exception of the Emu line. This way, the card sees that it's already at 48KHz and doesn't touch it. And believe me, there is a sonic difference; you don't even have to try. When I enabled this on my SB Live! (crappy card already), it was like the whole sound got a punch in clarity. Amazing.

 

[Zemo]

Quote:

Originally Posted by chia-pet I found this on a foobar2000 setup guide. Any interpretations?

Hehe....I wrote that guide!

-Z

 

[llmobll]

Quote:

Originally Posted by Stephonovich Dunno why no one's answered this for you yet... Resampling to 44.1KHz is never a good idea, unless you have really low source material and want to bump it up for some reason. The article snippet you posted referred to resampling to 48KHz will benefit you if you have a card that resamples to 48KHz internally, i.e. every card Creative's come out with, with the exception of the Emu line. This way, the card sees that it's already at 48KHz and doesn't touch it. And believe me, there is a sonic difference; you don't even have to try. When I enabled this on my SB Live! (crappy card already), it was like the whole sound got a punch in clarity. Amazing.

No one's mentioned it because those who are in the know all agree that resampling does little to nothing.

 

[Stephonovich]

Quote:

Originally Posted by llmobll No one's mentioned it because those who are in the know all agree that resampling does little to nothing.

Seems rather presumptuous of you... I know of quite a few people on this forum whos' opinions I trust who say resampling does quite a bit. Personally, I don't think it does much, but I leave it open to discussion. I also think your soundcard, and to a lesser degree, amp and headphones have a great deal on how much resampling will affect the sound.

All of this is meaningless anyway, as I'm referring to resampling to avoid dodgy hardware resamplers. It's been proven that Creative's resampler sucks bigtime, and that you WILL have an increase in SQ by avoiding it.

 

[llmobll]

You disagree with what i said yet agree. hehe


Sometimes i sound a little big headed, although I think that, especially on this forum, upsampling is sought after when in reality i haven't seen any personal SQ increases. It seems the hardware of the card, the buffers, opamps, and over design affect the sound more than the DAC.


I guess its because i'm back to 48K from 96K and I heard no difference. I'm waiting EMU to update their drivers and enable 192Khz playback to see if my assumptions are true. Unless someone can post some unbaised measurements or graphs all this talk is strictly just that... talk.

 

[Stephonovich]

Quote:

Originally Posted by llmobll You disagree with what i said yet agree. hehe

Erm, What? I said I don't think resampling does much, however, I believe it does a great deal with Creative cards via bypassing their internal resampler. Your original post, OTOH, seems to say that you don't believe any resampling will be beneficial.

 

[Zemo]

Lemme see if I can get this to simmer down a bit....

resampling does nothing to the sound quality. it only creates more samples. you can't "make water more waterey", or add quality that doesn't exist. The only way resampling will improve anything is if the hardware is designed to work with a certain samplerate. Feeding it that specific samplerate will allow the hardware to perform bettter, and viola! better sound! I've found that, since most hardware is designed to work at 44.1kHz, as that is the standard for redbook audio (aside from 10k2 based soundcards, like the creative audigy lines) 44.1, 88.2, or 176.4kHz sample rates tend to sound the best. Of course, this is all assuming a "perfect" resampling algorithm, which doesn't exist.

This is all from a playback standpoint, fiy. for recording, it's a different ballgame. (for more on recording and samplerates, good nyquist, and relevant theorys).

-Z

 

[llmobll]

Ok i think i'm being misquoted here to lets see if we can sort this out.


I do not here ANY difference when using any type of upsampling. period. I've tried it with many settings and i don't here ANY difference. Maybe when I get some diamond buffers for my PPA and recable my 595's i'l hear a difference. I doubt it.



When I said the whole "water more watery" thing I was talking with respect to using the resampler at 44khz, when the audio is at 44khz. Nothing is done, theoretically. SO there is no point in resampling to that specific rate.


Now if you disagree that upsampling makes a difference PLEASE tell me your rig specs aswell as how the sound is changed. Even though no one has posted any unbiased measurements to confirm that the sound is indeed different I'll believe someone when they tell me whole heartedly that they hear a difference. As with burn in i remain skeptical. You get a much bigger change in sound from the hardware's design than upsampling.