DAC vs. Stats -- FIGHT

[teufelshunde]

If you've got a recommend on a DAC that'll take a USB 24/192 feed, I'd love to hear your thoughts.

I like the sound of Sabre, but Wolfson chips would be OK too.

I'm willing to spend up to $2K.

Should have the new amp sometime next week, but won't have a DAC to feed it.

T.

Edited February 12, 2012 by n_maher
clean up ecp's thread

[jvlgato]

There's a lot of people here who know way more than I do about DACs. I've been very happy with the Perfect Wave DAC. A bit over your price range, but not too much if bought used, like I did. I like my Arcam rDAC for a budget DAC. The Lavry gets a lot of love around here, but I haven't heard it. The new NAD DAC looks very promising.

Doug is coming out with a DAC soon.

Lots of good options at that price, but you probably want to hear them before spending that kind of cash.

Enjoy the journey!

[teufelshunde]

Thanks for the suggestions John.

The Perfect Wave DAC reviews well and is currently available new at Music Direct for $1995 --> http://www.musicdirect.com/p-2456-ps-audio-perfectwave-dac.aspx

Issue is the USB input being limited to 24/96.

T.

[recstar24]

Might as well get Doug's dac.

[deepak]

Thanks for the suggestions John.

The Perfect Wave DAC reviews well and is currently available new at Music Direct for $1995 --> http://www.musicdire...ctwave-dac.aspx

Issue is the USB input being limited to 24/96.

T.

Any reason you have to have 24/192? I doubt anyone could statistically significantly tell the difference between 24/96 and 24/192

[crappyjones123]

Any reason you have to have 24/192? I doubt anyone could statistically significantly tell the difference between 24/96 and 24/192

i could, for alpha = 99.999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999%

i know its just the pedantic mathematician in me talking but it leaves this weird feeling in the back of my throat whenever people talk about statistical significance without mentioning the significance level.

Edited February 12, 2012 by crappyjones123

[grawk]

Definition for the pedant:

statistically significant: often enough to be greater than the odds of guessing at random

[crappyjones123]

Definition for the pedant: statistically significant: often enough to be greater than the odds of guessing at random

absolutely wrong and utterly misleading.

see the example under statistical significance.

http://www.stat.yale.edu/Courses/1997-98/101/sigtest.htm

any elementary stat book will explain why it is necessary to give a significance level or a p value otherwise the test is meaningless.

[grawk]

note: this is not a statistics textbook, it's a conversation on a web forum.

[crappyjones123]

"hmm...crappy is right but i can't let that pass. so i am going to backtrack and say that the preciseness doesn't matter even though i tried to correct him on it (and failed)"

just admit you were wrong instead of trying to safe face.

sorry to thread crap doug but needed to get that out.

[morphsci]

Actually a significance level, or more correctly an alpha level (the type I error rate) is not actually necessary. One can simply calculate the probability of the test and utilize it in the context of the original hypothesis without the need of a totally arbitrary alpha level.

[grawk]

I wasn't giving you a textbook definition, I was giving you a contextual definition so you could pull your head out of your mathematical ass

[morphsci]

He actually had it in sideways anyway.

[crappyjones123]

im assuming you are talking about the probability of a type 1 error. how does one define low, even in the context of the null hypothesis? surely, the drug companies would like low to be 10% so they can peddle their wares but would you prefer to take a drug that the fda cleared for consumption at alpha = 1% or alpha = 10%. "low" could mean different things depending on who is looking at things.

http://www.stat.columbia.edu/~gelman/research/published/signif4.pdf

how does one provide confidence intervals without alpha being predefined?

[morphsci]

The level of significance is arbitrary because as you stated, it can be any value. Therefore you simply calculate the p value, which is the actual probability of making a Type I error. The whole idea of a significance level is archaic. It's only usefulness is in introductory classes and to explain what an actual probability value of a test statistic is. Confidence levels are also arbitrary as you can calculate them for any level of confidence between 0 and 1. At least in biology we are rapidly moving away from the idea of statistical significance. The Bayesians have given up on it a long time ago.

Now back to your regularly scheduled program.

[luvdunhill]

note: this is not a statistics textbook, it's a conversation on a web forum.

Note: this isn't a web forum but a group of friends hanging out at their favorite neighborhood bar. No need to be an ass.

[Beefy]

At least in biology we are rapidly moving away from the idea of statistical significance. The Bayesians have given up on it a long time ago.

I'll give you my p<0.05 ANOVA when you take it from my cold, dead hands.

[teufelshunde]

Any reason you have to have 24/192? I doubt anyone could statistically significantly tell the difference between 24/96 and 24/192

I don't have to have 24/192, but have you noticed how strongly correlated it is with the presence of a particularly pleasing psycho-acoustic placebo effect?

[crappyjones123]

didn't know biology was moving away from it. it is still the back bone of model testing for robustness and risk analysis. at least for the things i worked on, i couldn't go 5 pages without confidence intervals making an appearance.

whacha coin with anova beefy?

[jvlgato]

I believe the PWD Mk2 has 24/192 via USB. But the Mk2 costs more, putting it it of your price range. You could always upgrade later, if you felt it necessary. I've actually never heard 24/192, so don't really have an opinion. I think you should have a listen to a handful of DACses that seem to fit the bill, and you'll just know what's right for you.

http://www.psaudio.com/products/audio/media-players/perfectwave-dac/

Edited February 12, 2012 by jvlgato

[jvlgato]

To my knowledge, p<.05 in a drug study is still the accepted standard for a drug's effect to be likely enough to not be due to chance, and therefore be acceptable data to evaluate for a drug's efficacy. Arbitrarily chosen, but still used.

[atothex]

I don't even have anything in 24/192. I must be like behind the times or something, but where do you get that?

[Dreadhead]

Well through this argument I have read more about hypothesis testing than I had ever before and I think I am going to have to learn some more for an upcoming paper. Any suggestions as to reading? I am assuming wikipedia probably isn't the best.

[crappyjones123]

Lehmann is a typical grad text.

http://www.amazon.com/gp/aw/d/0387949194

I'll see if I still have my copy. It had a nice rigorous treatment.

[jvlgato]

I don't even have anything in 24/192. I must be like behind the times or something, but where do you get that?

I have seen it in HDTracks. But since I can't play it, I never bothered. I do like 24/96, though! I suppose if I could play it, I'd probably download a track out of curiosity. But I've been happy enough with 16/44.1 and 24/96 through the PWD for my main rig and recently, though the Arcam rDAC for my bedroom rig, that I've never had the itch.

Teufelshunde, what have you heard in 24/192, and what did you notice about it?

Edited February 12, 2012 by jvlgato