Originally Posted by audiophilesavant
With noise shaping, we have dynamic range which exceeds the threshold of human hearing with 16 bits. You have posted charts which demonstrate that. Professional recording engineers record using DSD or 24 bit PCM. Professional recording engineers also use noise shaping when they dither down from DSD/24 bits to 16 bits when producing CDs. CDs are therefore capable of delivering all the dynamic range that is required by human hearing.
Who says "professional engineers" are using noise shaping? You have spectrum analysis showing that vast majority of titles have noise shaping? Or are you asking us to take your word for it?
Here is the UI for conversion of the track we are talking about to 16/44.1:
You see the default I have circled? No noise shaping? You think the typical creative type would naturally gravitate toward overriding the default and selecting something called "noise shaping?" How many do you think know what noise shaping is? Why didn't Adobe make that the default?
But sure, let's have you show us a sampling of spectrum analysis that you have done yourself to arrive at that conclusion or market research that indicates noise shaping is common in production of CD.
Do I think 120db of dynamic range is required for playback in our listening rooms? No. The background noise in the recording venue, the actual dynamic range of the music, the noise in the recording equipment from microphone to hard disk (or worse yet analog tape), the noise in the mixing and mastering equipment, which often includes multiple digital to analog to digital processes, the noise in the playback equipment, and the background noise in even the best listening rooms all mitigate against it.
All of that layman logic has been debunked with the research papers I presented earlier: http://www.avsforum.com/avs-vb/showt...5#post22061565
The confusion consistently comes from assuming noise is noise. As Arny rightfully said, numbers can just be numbers. You have to correlate them to what we hear. And what we hear is not a flat frequency response. So single number noise dB means nothing. This includes the analysis we just did for the audio track, and your read of amplifier S/N. Do you know what a-weighting is? It is a simplified model overlaid on measurements to make them somewhat realistic relative to what we actually hear. This is what it looks like:
You see that shape? Now look at the threashold of hearing:
You see how the a-weighting takes down the amplitude of noise in low frequencies very similarly to where the ear is least sensitive?
I have even quoted the specific sections from research that talks about all of this so you don't even need AES subscription to read it or ability to find the right section and the point is still missed. Yet folks keep talking about "oh, but do we hear it?" Well to know that you better use psychoacoustics and not blind metrics devoid of that. And don't substitute your common sense for top research that has already done this.
Proper analysis of the high-res track we just did would have to analyze the spectrum of background noise and determine how much of it is in the mid-frequencies. That is what matters the most. The dynamic range there may be far higher. Simple ultrasonic (> 20 Khz) noise could have set the noise floor for us yet we know such things are not audible and there must be excluded. But the dumb analysis tool we have can't do that. This is why I put all the caveats there. Please don't confuse anecdotal info with real research data.
What is the actual dynamic range required for a playback medium. That is determined by the levels of background noise in the recording and playback environments, the music itself, and the weakest link in the recording and playback chain. It varies from recording to recording, and system to system, but it is nowhere near 120db, and even if it were, CDs are capable of delivering it.
Well, you are clearly mistaken based on data in front of you. Go and read the Fielder papers and tell me where he didn't know what he was doing when he measured a) peak levels in live venues and recording equipment and b) how quiet ordinary and professional listening rooms are.
The file you analyzed and the file of the same recording Arny analyzed were both 24 bit; therefore, although they were of different sampling rates, they both have the same theoretical dynamic range (without noise shaping) of 144db. (Why HDtracks downsampled it to 96k instead of 88.2k is beyond me. Okay, it was for marketing purposes). Arny determined, and you have now agreed, that the actual dynamic range of the recording is in the range of 70db.
No, I didn't agree with that. You didn't understand the technical explanation I provided not once, but twice. Audiotion reported *RMS* values. Those are averages. The DAC has to represent real voltage values at every instance in time, not an average value. That means peak values. You must back out the conversion from peak to RMS that Audition performed. Doing that gets us to 96 dB for sine wave which is he default in Audition. Arny says he has material that is quieter than that. Therefore with the data in front of us as of this moment, there is nothing that says 70 dB is enough. This is of course on top of other reasons why this is not the number to be trusted.
It was recorded by one of the acknowledged master recording engineers of all time, one, who unlike most others, does not employ compression, which would have further reduced the dynamic range of the recording. Yet it is 50db shy of 120db. Although Arny alleges that he knows of recordings with greater dynamic range, I think you would be hard pressed to find them.
I would be? How would you know that? Based on what data? Gut feeling? Are you an industry expert with substantial experience in this area so we should trust your gut? And Keith released the tracks as 24 bit for what reason if he knows what he is doing and all that was there was 70 dB?
If audiophile recordings made by master recording engineers only have a dynamic range of 70db, I think the hand-wringing over the necessity of 120db is misplaced.
Let's be clear: the 120 dB did NOT come from me. It came from top experts in the industry who presented papers at Audio Engineering Society based on real research and measurements. Maybe you have a record of them being laughed out of the room when they knew the reality was 70 dB but I don't. Here is the quote again from the research:"Dynamic-Range Issues in the Modern Digital Audio Environment
The peak sound levels of music performances are combined with the audibility of noise in sound reproduction circumstances to yield a dynamic-range criterion for noise-free reproduction of music. This criterion is then examined in light of limitations due to microphones, analog-to-digital conversion, digital audio storage, low-bit-rate coders, digital-to-analog conversion, and loudspeakers. A dynamic range of over 120 dB is found to be necessary in the most demanding circumstances, requiring the reproduction of sound levels of up to 129 dB SPL. Present audio systems are shown to be challenged to yield these values."
In your next post, please don't address me. Address Fielder. Give us your credentials and body of data that would say he is mistaken.
What is the level of noise in your reference theater, which, having been designed by Keith Yates, is undoubtedly among the best listening rooms available? Subtract that number from 120db and let me know what you get.
If I did that, it would show that you don't understand psychoacoustics. Would you really like me to go there?
Here is the measurements from Fielder of consumer
You see that someone has built a quieter room than threshold of hearing. Importantly note how he has the spectrum of the noise and he compares it one for one against threshold of hearing. You cannot trust a meter with a dumb number thinking it means the audible
noise level of the room. You would be ignoring the way our brain and ear work together. A rumble from a freeway five miles away at 18 Hz would register on that meter yet can be totally inaudible. The fact that it shows up at -60 db means absolutely nothing.
Let's cap this off with this: http://www.stereophile.com/content/m...r-measurements"The No.532H is also superbly quiet; its wideband, unweighted signal/noise ratio, ref. 2.83V output with the input short-circuited, measured 93dB in both channels. (The A-weighted ratio was 105.4dB.) With its high output power, the No.532H is one of the few amplifiers that can match the dynamic range of true high-resolution recordings."
You see the a-weighting? You see how much higher it is than none, which at 93 db is 8 db higher than your assumed "85 dB?" You talk about spins. You think you did OK by changing a "> 85 dB" to 85 dB and calling it done?