[frankR]

Okay.

So reinstalled ASIO4ALL. Before the last BSOD I was still having the bit drop-out popping.

But now it seems to have sorted itself out, I'm now running 192khz!! The device selected in the output is not ASIO4ALL, it's ASIO 2.0. But it's all working well, using Latency of 1024.

It sounds spectacular! It's hard to imaging it sounding any better, can't wait to try a deticated headamp. Recordings that I thought were bad before my HD650s/Audiotrak Prodigy HD2 sounds surprisingly good.

All the background sounds effects on the Mars Volta album Bedlum in Galiath are prominent, it's quiet satisfying.

Thanks again, B0dhi.

Edit: I've been hearing some popping. But the popping repeats, so they must be bit drop-out from the CD rip or flaws in the recording, even though this CD was ripped straight from the shrink-wrap.

 

[Andrew_WOT]

Quote:

Originally Posted by frankR /img/forum/go_quote.gif I found the secret rabbit code, "code" literally. It comes uncompilled. I don't have a C++ compiler on my machine. It would be nice if there was compiled version out there.

Secret Rabbit Code Resampler for Foobar2000.

 

[sschmeichel]

I downloaded the WINAMP5用 ASIO出力プラグイン (dll version) - ãŠãŸã¡ã‚ƒã‚“ã®MIDI/Audioソフト.

But how can I use it? rename it to a dll file?

 

[frankR]

Quote:

Originally Posted by Andrew_WOT /img/forum/go_quote.gif Secret Rabbit Code Resampler for Foobar2000.

That secret rabbit code plugin for foobar seems to be more processor intensive then other codes I've used. I can't make anything above 96khz play, at 196khz it nearly maxes out one of my processors.

 

[gregorio]

Just out of curiosity, what are you people doing?

Let me make a few points clear:

1. Upsampling cannot improve the quality of the audio. A CD is sampled at 44.1kFs/s and by definition cannot contain any frequencies above 22,050Hz. Upsampling cannot magically find the audio frequencies which have been deliberately, totally and irrevokably removed from the original recording. If you do find frequencies beyond 22050Hz when you've upsampled I would return the DAC and ask for your money back because it's malfunctioning!

2. If you think your audio sounds better at a higher sample frequency than the original 44.1kFs/s, what you are really hearing is a smoother reconstruction filter in the DAC, not an improvement in the quality of the audio file itself. This is telling you that you have a cheap and nasty DAC whose implimentation of a reconstruction filter at 44.1k is so poor that anything else sounds better.

3. Higher sample frequencies make sense only for poor quality DACs (or ADCs for that matter). It is difficult to make an anti-alias filter or reconstruction filter work without artefacts over a small range of audio frequencies. Higher sample frequencies allow for smoother, more easily implemented filters. If you can hear a difference between 44.1kFs/s and say 96kFs/s, it's the implementation of these filters you can hear.

Last time, it is not humanly possible for you to hear a difference between a 44.1kFs/s audio file and an audio file with a sample rate of 88.2, 96, 176.4 or 192kFs/s. There are NO exceptions to this rule, providing of course that you are a homo-sapien!!

Gregorio

 

[frankR]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif Let me make a few points clear: 1. Upsampling cannot improve the quality of the audio. A CD is sampled at 44.1kFs/s and by definition cannot contain any frequencies above 22,050Hz.

Allow me to revise your statement: Upsampling can not increase the Nyquist frequency (half sampling frequency) of the source recording. However I believe it can improve sound quality through interpolation which results and a smoother more analog reproduction of the wave form.

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif This is telling you that you have a cheap and nasty DAC whose implimentation of a reconstruction filter at 44.1k is so poor that anything else sounds better.

The AKM4396 must be a poor DAC because I believe the upsampled audio sounds better to my ears. High frequency sounds like symbols sound more natural.



Quote:

Originally Posted by gregorio /img/forum/go_quote.gif There are NO exceptions to this rule, providing of course that you are a homo-sapien!!

I guess I’m some evolved species of human then.

 

[Brewmaster]

It seems that many premium stand alone DAC's upsample in hardware before feeding the signal to the DAC chip. This is even with top of the line DAC chips.

Therefor I suspect there is some merit to doing this. I've never gone down this road myself, but it may improve sound quality.

 

[gregorio]

FrankR - "AKM4396 must be a poor DAC", hurah, sounds like you're starting to get it. The AKM4396 might be a good DAC in other respects but if you say you can hear a difference then indeed it's 44.1kFs/s reconstruction filter must be pretty ropey. BTW, what makes you think interpolation gives a smoother more accurate waveform? If anything it creates quantisation errors, it doesn't remove them. I don't know where you got this bit of information from but it's incorrect.

There is no decent professional DAC that would upsample, in fact I can't even think of a cheap professional DAC that would upsample. Oversampling certainly but not upsampling. What professional could afford to potentially monitor quantisation errors which aren't present in the original digital audio files.

Brewmaster - I'm not sure why top of the line (consumer) DAC chips are automatically upsampling, maybe it's cheaper to upsample and put it through a 96k reconstruction filter than just to design a decent 44.1k reconstruction filter in the first place. This is certainly not the route that professional DACs take. Sounds to me like a ploy to maximise profits in the consumer marketplace by taking advantage of the fact that consumers are not going to know much about digital theory!

Gregorio

 

[frankR]

Gregorio,

I'll stop debating you about the merits of upsampling because the difference are so subtle they're almost inperceptable. However, for whatever the reason, I believe the upsampled audio sounds better.

However, I am interested to hear your response to my frequency spectrum analysis of the 16/44 and 24/96 audio.

Let me ask you something. Do you believe upsampling images has merit?

 

[gyrodec]

gregorio

Yesterday you said "... than just to design a decent 44.1k reconstruction filter in the first place", as part of your rant against upsampling. However, I would like to suggest that it impossible, within the laws of physics we are all bound by, to "build a decent 44.1k reconstruction filter". All steep filters have serious issues, usually down to fs/10 and usually in the phase response. So, moving the filter point will always be a good idea. The only question is, can you do the resampling with fewer errors than are cause by the lower filtering point. Thats just an engineering trade-off, there is no right or wrong answer, just good and bad implementations. (I agree pro audio shouldn't resample for the recording chain, but I see no reasons why they shouldn't do it in the monitoring chain.)

 

[gregorio]

Gyrodec - Commercial recording studios spend 100,000s to ensure that what they are monitoring is as exactly as possible what they are recording.

All filters have issues, steep ones just have more of them. There have been units which managed to impliment great sounding filters at 16/44.1. I remember one 16/44.1 DAC a few years ago which easily exceeded the quality of most 24/96 pro-gear on the market, unfortunately I also remember it cost well over $10,000.

At the end of the day the DAC is less than half the story. What's more important is the ADC, afterall, it's the ADC which implements the anti-alias filter in the first place. Even a perfect DAC can only represent what's been created by the ADCs. A high end pro ADC running at 16\44.1 will out perform an affordable pro ADC running at 24\96.

 

[b0dhi]

As gyrodec explained, 44.1 is inherently worse than 96Khz or higher sampling rates. The only reason studios still mix in 24/44.1 is because of limited processing power (and the fact that it will be mixed down to 16/44.1 in the end anyway). They will multitrack in 16/44.1 for the same reason. Recording is still done at the highest possible (affordable) sampling rate and bit depth.

 

[gyrodec]

Don't confuse the sound of a better filter with the sound of a better power supply/analog stage and a more stable/accurate clock, these two things are usually the cause of the differences between the more expensive gear, and can easily outweigh the differences in DAC filter point.

I do agree that the ADC filter is the deal breaker, and I would much rather mine be at 88.1/96/179.2/192 and not 44.1 as those brick-wall filters are real pices of work.

 

[gregorio]

b0dhi - 44.1 is not inherently worse that 96k, where on earth did you get that piece of information from? Likewise it is not true that studios only record at 24/44.1 because they don't have the processing power to do 96k. Many studios still use 44.1kFs/s because the process of resampling to 44.1k from 96k is more destructive than the damage done by the anti-alias filters in high quality ADCs.

Gyrodec - It depends on the ADC you are using, a top quality ADC's filters at 44.1kFs/s are virtually indistinguishable from the filters at higher sample rates.

 

[sejarzo]

Wouldn't it make sense, then, for studios to record at 24/88.2.....better resolution and far easier to convert to Red Book?