Thread: Audio advisor suggests using PC as head unit is horrible...

Originally Posted by smellygas

97dB is an EXCELLENT SNR for 16-bit 44.1kHz, which is the audio format of CD's and most mp3's, and just about approaches the limit possible with 16-bit recording!!! What are you talking about!!??

I'm talking about the hardware cababilities as it relates to digital audio resampling, regardless of the media sample rate...listening to 16 bit/44Khz audio upsampled to 24/96Khz will always sound better than 16 bit/44Khz output...

Originally Posted by smellygas

Your Xonar's SNR is only 119dB with 24-bit recording and playback. It is actually 96dB with 16-bit 44.1kHz, which is the relevant number, unless happen to be listening to 24-bit recordings in your CAR.
http://audio.rightmark.org/downloads...tGuide_V12.pdf

Yes, I have DVD-A and SACD media in my car!
Again, it's not the sampling limitations of the media, but the cababilities of your hardware that matters, and at what rate your outputing...
If I have a card thats 32bit/192Khz capable, Im certainly not going to limit it to 16/44Khz! Especially since you kindly provided proof that the higher sample rates are superior as far as SNR, crosstalk and dynamic range on my Xonar...

Originally Posted by smellygas

Incorrect. 3dB doubles the POWER (e.g. watts), but not the perceived LOUDNESS. 20dB is NOT twice as loud as 17dB. A doubling of perceived loudness requires anywhere from 6-10dB increase.

Simply not exactly true...a decibel can represent POWER or intensity, but it can also be used to describe amplitude, voltage or even current.
But as it relates to this discussion (accoustics), we're talking about sound level...
And sound level is a measure of PRESSURE...
"The human perception of, for example, sound or light, is, roughly speaking, such that a doubling of actual intensity causes perceived intensity to always increase by the same amount, irrespective of the original level. The decibel's logarithmic scale, in which a doubling of power or intensity always causes an increase of approximately 3 dB, corresponds to this perception"
So, you see, as decibels relates to sound, it's all based of the preception of the listener. 3dB doubles the intensity...not 6 to 10dB!

Originally Posted by smellygas

WuNgUn, once again, there is a lot of misinformation if your post. If you're not entirely sure about a topic, perhaps you shouldn't post about it.
SG

So tell me again, what my misinformation is, 'cause you failed to convince me with your post...

Originally Posted by WuNgUn

I'm talking about the hardware cababilities as it relates to digital audio resampling, regardless of the media sample rate...listening to 16 bit/44Khz audio upsampled to 24/96Khz will always sound better than 16 bit/44Khz output...

WRONG. If you take a 16-bit/44.1kHz recording which has absolute limit of ~96dB and you upsample it to 24-bits and 96khz, assuming a PERFECT conversion, guess what your new SNR is? It's 96dB!!!!!!!! Furthermore, whenever you manipulate or alter a signal, it causes DEGRADATION. I don't know how you could even think that changing the bit depth and extrapolating data that doesn't exist to get 96kHz sampling could possibly result in BETTER sound.

Again, it's not the sampling limitations of the media, but the cababilities of your hardware that matters, and at what rate your outputing...
If I have a card thats 32bit/192Khz capable, Im certainly not going to limit it to 16/44Khz!

Just about everybody here is playing mp3's and CD's. That's 16-bit/44.1kHz, my friend. Even if the DAC on your sound card supports up to 32-bit/192kHz, it is IRRELEVANT because when you're playing back mp3's and CD's, your card is operating in 16-bit/44.1kHz mode.

Simply not exactly true...a decibel can represent POWER or intensity, but it can also be used to describe amplitude, voltage or even current.
But as it relates to this discussion (accoustics), we're talking about sound level...
And sound level is a measure of PRESSURE...
"The human perception of, for example, sound or light, is, roughly speaking, such that a doubling of actual intensity causes perceived intensity to always increase by the same amount, irrespective of the original level. The decibel's logarithmic scale, in which a doubling of power or intensity always causes an increase of approximately 3 dB, corresponds to this perception"
So, you see, as decibels relates to sound, it's all based of the preception of the listener. 3dB doubles the intensity...not 6 to 10dB!

You misinterpreted your own quote. It is saying that humans perceive a doubling (exponential) of intensity by the same amount (linearly). Your quote does not say that perceived loudness doubles when sound intensity doubles. There's a big difference.
This table should make it REALLY easy for you to sort this out:
http://www.gcaudio.com/resources/how...eloudness.html

dB Change, Voltage, Power, Loudness
3 1.4X 2X 1.23X
6 2.0 4.0 1.52
10 3.16 10 2

SG

Originally Posted by smellygas

WRONG. If you take a 16-bit/44.1kHz recording which has absolute limit of ~96dB and you upsample it to 24-bits and 96khz, assuming a PERFECT conversion, guess what your new SNR is? It's 96dB!!!!!!!! Furthermore, whenever you manipulate or alter a signal, it causes DEGRADATION. I don't know how you could even think that changing the bit depth and extrapolating data that doesn't exist to get 96kHz sampling could possibly result in BETTER sound.

That very link you provided of the RMAA of the Xonar shows how the SNR improves as you increase the sampling, all the while using a 16 bit or 24 pass thru signal...
The ONLY thing changing was the sampling, and the SNR improved. Explain this to me?
It's called upsampling...
And there is no alteration or manipulation to the signal!! It's strictly passive...it's reading, or SAMPLING the digital data at a higher rate. It's simply interpreting the SAME data at a higher rate...

And so I guess you also believe that resampling 16/44 audio to 24/96 has no benefit other than taking up more space??

Originally Posted by smellygas

Just about everybody here is playing mp3's and CD's. That's 16-bit/44.1kHz, my friend. Even if the DAC on your sound card supports up to 32-bit/192kHz, it is IRRELEVANT because when you're playing back mp3's and CD's, your card is operating in 16-bit/44.1kHz mode.

"A higher resolution DAC, for example, 24-bit is of higher resolution than 16-bit, improves the sound quality of audio playback by improving the overall signal-to-noise ratio (SNR) of the analog output"

Anyway...

Originally Posted by WuNgUn

That very link you provided of the RMAA of the Xonar shows how the SNR improves as you increase the sampling, all the while using a 16 bit or 24 pass thru signal...The ONLY thing changing was the sampling, and the SNR improved. Explain this to me?

No, no, and NO. Those were RMAA loopback test results at different bitdepth/samplerates. Line-out to line-in. For the 16-bit/44.1kHz test, it takes a 16-bit/44.1kHz file, converts it to analog to line out, then samples it at 16-bit/44.1kHz from line in (via loopback cable), and it gives you a measurement. For the 24-bit/96kHz test, you start with a 24-bit/96kHz signal, convert it to analog, and sample it at 24-bit/96khz. Each test is independent. NO CONVERSION FROM 16-BIT TO 24-BIT OR 44.1KhZ TO 96KHZ IS OCCURING = RMAA is NOT a test of resampling. You can't just look at a table of numbers and make up a story to explain them. You have to understand what they're referring to.

It's called upsampling...And there is no alteration or manipulation to the signal!! It's strictly passive...it's reading, or SAMPLING the digital data at a higher rate. It's simply interpreting the SAME data at a higher rate...

Oh is that right? Let's see. So if you have a 44.1kHz signal, it means that you have an amplitude value every 1/44100th of a second. However, if you wish to "resample" to 96kHz, you're expected to have an amplitude value every 1/96000th of a second. So if you only 44100 data points in a second, how on earth are you going to come up with 96000 data points for your new sample rate!? Are you just going to make up data for the missing 51900 samples? Seriously, man.

And so I guess you also believe that resampling 16/44 audio to 24/96 has no benefit other than taking up more space??

NONE. Unless you have a specific reason, like you're doing mixing with other 24/96 recordings...or if you're doing audio editing.

"A higher resolution DAC, for example, 24-bit is of higher resolution than 16-bit, improves the sound quality of audio playback by improving the overall signal-to-noise ratio (SNR) of the analog output"

Anyway...[/QUOTE]

That's a no-brainer quote. Everybody knows that 24-bits has a lower noise floor than 16-bit, BY DEFINITION. A 16-bit signal is incapable of providing the SNR of a 24-bit recording because the SNR is limited by the # of bits!! It doesn't matter how many bits your DAC chip is capable of.

SG

Maybe you should read the first couple of chapters of this...
http://photos.imageevent.com/cics/v0...rts%20v0.3.pdf
...and give me your expert opinion on resampling then....

Originally Posted by WuNgUn

Maybe you should read the first couple of chapters of this...
http://photos.imageevent.com/cics/v0...rts%20v0.3.pdf
...and give me your expert opinion on resampling then....

Interesting paper on one way to recreate a high-end transport using your computer. I did read the section on upsampling, and to be honest, it really doesn't explain why the sound should be any better. Here's why. Sure, you get a smoother waveform when you interpolate values in between the original samples (note that this is not a "passive" process), at the expense of generating artifacts...however, most cd players and DAC's ALREADY do this as part of the oversampling process...(or the equivalent of oversampling in certain DAC topologies). I personally hate literature published by audio companies because they're usually filled with nonsense, but I think this explanation just about sums it up: http://www.thetadigital.com/upsampling.htm

You would have to have a very very good upsampling algorithm as well as a high quality DAC that does well at the higher bit depth and sample rate in order NOT to comprise your sound quality when you resample. I'd be interested in seeing some blinded comparisons.

SG

I'm also kinda put off on the idea of 'filling in the blanks' of the original data...just doesn't seem right!
But a smoother waveform does represents something closer to an analog wave. I think you run the risk of more artifacts when there isn't enough data between samples, like you say, most CD players and DAC already do this...
The 'Secret Rabbit Code' seems to do it better than most...

Resampling from 16 bit to 24 bit will not improve your SNR or give you a better SQ. All you are doing is finding the 24 bit equivalent of the 16 bit signal. the increased sampling rate only oversamples the original sample and the tries to estimate the intermediate values to the next sampled value. The is accomplished by adding additional bits. So although you end with a a smoother slope the signal is no more accurate than the original 16 bit signal. It's like taking a 700MB avi file and using it to make 3.5 GB DVD. you have 5 times as much data on the DVD as the original AVI but the quality is about the same. Sometimes the resulting DVD quality is degraded (artifacts, lower SNR).