Is the dynamic range of CD sufficient?

[JackD201]

Jack,

But excessively simple IMHO. There is more than Xmax and Xmin .When you establish the dynamic range of a concert people talk of the ratio between the maximum level and the noise level of the room . But the noise the room is useful information and part of the recording and must be encoded - it "needs some bits" - I do not know how many. The classical definition of dynamics compares the maximum level of undistorted signal with noise from electronics - that has a random behavior and does not represent any useful information. IMHO we can not go from one to the other just directly. It is why some people say that at less 18 or 20 bits are needed for subjective quality listening at typical sound levels, adding two extra bits for implementation losses, making it 20 or 22.

The preference for an higher number of bits, or in the case of Mark , DSD with its higher dynamic range is a clear subjective indication that the low levels are not clearly reproduced in 44.1/16. Our brain prefers the representation that has less errors.

It may seem excessively simple my friend but that is where it begins. That in tandem with the highest and lowest frequencies. It is these parameters that determine the selection of settings. In the former the the recordist will adjust gain if his bit depth is limited and the latter the LPF/AAF if his sampling rate is limited. The majority of the time if not all the time the first limitation is the transducer used. As you know selection is based on composite rather than a pure tone. That's where I see a lot of people getting hung up on sampling theory. They see a reconstructed pure sine wave and say its perfect there ergo a composite signal is perfect. To me that is giving the scientists and engineers that developed digital in the first place and continue to attempt to improve it, short thrift. The math being solid has nothing to do with the performance the degree at which a device under test actually living up to the math. That is two separate issues.

The question is, how many quantizations is enough to give the semblance of a continuous COMPOSITE wave form. The answer depends on how faithful you want it. If it is merely for passable performance, many would be surprised how well 12bit does. Going upwards, talk about diminishing returns because what is being chased is infinity.

So where does one draw the line? In studios people are now starting to record at 32bit. Processing today is being done with machines running 64bit floating point. DACs today use 32bit engines (more to be able to do digital volume control and for more DSP). These people aren't doing it just because they can. There are very clear, very real benefits.

Like you I'm invested in quality CD playback because like you I have music in that format that I love. That is the only reason. While I have no intention of doing my entire CD library at a higher bit depth, sampling rate, I do know the comparative superiority of the same over the 16bit/44.1 standard. Now that is an important word. Standard. It wasn't Nyquist or Shannon. It was Sony and Philips who set it. One need not go far to find why they chose it. That was what was practical and economical AT THAT TIME. Way good enough to beat a cassette tape or a mass market table with an MM cart set up by someone who didn't own so much as a cardboard alignment template playing beat up records.

Well, the world tends to move on. 24bit is now economical and practical too. While many people try to shoot it down as excessive, they should be glad that some producers are even making it available given how vulnerable they can become in this world of modern piracy.

 

[marty]

I found the discussion of dynamic range on a CD interesting. I realize the discussion centered on whether it was adequate to reproduce a concert-like experience at home. But I'd like to ask a slightly different question which is something that I suspect many of you know. Very simply, what has the better dynamic range, a CD or an LP? Let me put this into a pragmatic context. If I were to listen to an LP and a CD made from the same master tape, and I was listening to a pppppp passage and set the levels to be similar, which would sound louder during subsequent identical ffff passages, the CD or the LP? Perhaps the answer varies due to mastering skills, but I'm hoping not to get stuck in technical minutia, as important as they may be. Rather, let's say we took 6 RCA Living Stereos titles and compared the LPs to their respective red book CDs, on average, would we find one to have superior dynamic range compared to the other as a general rule?

 

[esldude]

I found the discussion of dynamic range on a CD interesting. I realize the discussion centered on whether it was adequate to reproduce a concert-like experience at home. But I'd like to ask a slightly different question which is something that I suspect many of you know. Very simply, what has the better dynamic range, a CD or an LP? Let me put this into a pragmatic context. If I were to listen to an LP and a CD made from the same master tape, and I was listening to a pppppp passage and set the levels to be similar, which would sound louder during subsequent identical ffff passages, the CD or the LP? Perhaps the answer varies due to mastering skills, but I'm hoping not to get stuck in technical minutia, as important as they may be. Rather, let's say we took 6 RCA Living Stereos titles and compared the LPs to their respective red book CDs, on average, would we find one to have superior dynamic range compared to the other as a general rule?

CD has the ability to have more dynamic range on playback. But mastering is going to make a difference perhaps. LP's always have to be EQ'd extensively for a variety of reasons going onto the stamper disc. Then EQ'd back the other way on playback. The ability of cartridges to track enter into it as well. Bass is often summed etc. etc. If you used 6 Living Stereo CD's and LP's they aren't going to be the same.

Some friends and I years back did a telling comparison. We had a couple reel-to-reel collectors in our group, all owned records and CD's. We managed to put together a sizable number of recordings that we had on all three formats. As close to the same as we could get. The reels were all pre-recorded reels. One thing became obvious. The LP, each and every time was the odd man out. Now LP it might sound better to us sometimes (though not the majority), but the CD's and reels were far closer in general balance. Even with tape hiss the general character was more similar between CD and pre-recorded reels. That was a big surprise. All of us were thinking CD would be the odd man out or perhaps they all would be rather different. We had a Studer Revox reel, and comparable equipment for the times for the other two formats. Speakers were Quad ESL 63 and Thiel and Magnapan as we did this more than once at more than one location. Same general results each time.

Sorry, I couldn't simplify it more.

 

[Al M.]

CD has the ability to have more dynamic range on playback. But mastering is going to make a difference perhaps. LP's always have to be EQ'd extensively for a variety of reasons going onto the stamper disc. Then EQ'd back the other way on playback. The ability of cartridges to track enter into it as well. Bass is often summed etc. etc. If you used 6 Living Stereo CD's and LP's they aren't going to be the same.

Some friends and I years back did a telling comparison. We had a couple reel-to-reel collectors in our group, all owned records and CD's. We managed to put together a sizable number of recordings that we had on all three formats. As close to the same as we could get. The reels were all pre-recorded reels. One thing became obvious. The LP, each and every time was the odd man out. Now LP it might sound better to us sometimes (though not the majority), but the CD's and reels were far closer in general balance. Even with tape hiss the general character was more similar between CD and pre-recorded reels. That was a big surprise. All of us were thinking CD would be the odd man out or perhaps they all would be rather different. We had a Studer Revox reel, and comparable equipment for the times for the other two formats. Speakers were Quad ESL 63 and Thiel and Magnapan as we did this more than once at more than one location. Same general results each time.

Sorry, I couldn't simplify it more.

That is what many recording engineers have said when asked about CD vs. LP, the CD is closer to the source tape in general balance and character.

 

[Al M.]

That is what many recording engineers have said when asked about CD vs. LP, the CD is closer to the source tape in general balance and character.

Having said that, I also have found myself preferring LP over CD in some circumstances. A friend of mine has a very good turntable and a reasonably good CD system, better than average. In his set-up, I strongly and consistently prefer LP -- the CD playback is clearly weaker and less involving in sound.

However, when I compare my CD playback at home, I find it easily as good sounding as his LP system, though again, different in character, and at the same time closer to the real thing (when it comes to live unamplified music).

I am always wondering about audiophiles who have little experience with going to concerts of unamplified live music (e.g., classical) and mostly listen to rock and pop: what does it really mean when they prefer LP over CD? I guess it says more about their tastes, which I have sympathy for, and perhaps about the relative qualities of their LP vs. CD playback systems, rather than it says anything about a reasonable comparison with a valid reference. What does it mean when they say "LP is better"?

 

[Julf]

In studios people are now starting to record at 32bit.

One has to be careful not to confuse 24 bit fixed-point PCM and 32-bit floating point. 32-bit floating point actually has less precision than 24-bit fixed point.

Processing today is being done with machines running 64bit floating point. DACs today use 32bit engines (more to be able to do digital volume control and for more DSP). These people aren't doing it just because they can. There are very clear, very real benefits.

Yes. For processing, where you need headroom and extra precision for DSP operations. That is different from storage/transmission/replay.

 

[JackD201]

I know Just pointing out that the processing power is already in the field.

 

[Al M.]

Yes. For processing, where you need headroom and extra precision for DSP operations. That is different from storage/transmission/replay.

Yes, as the article says that I quoted in my OP:

While professional editing, mixing, processing, equalizing and level shifting usually use more data bits for computation (24 bits linear, 32-bit floating point or now 48-bit linear), 16 bits is more than enough for unlimited fidelity as a release format.

The reason we use more bits in production is so we can create and preserve a true 16 bits through the whole process after all the truncation and rounding and nastier stuff that goes on between the microphone and your CD.

 

[microstrip]

Having said that, I also have found myself preferring LP over CD in some circumstances. A friend of mine has a very good turntable and a reasonably good CD system, better than average. In his set-up, I strongly and consistently prefer LP -- the CD playback is clearly weaker and less involving in sound.

However, when I compare my CD playback at home, I find it easily as good sounding as his LP system, though again, different in character, and at the same time closer to the real thing (when it comes to live unamplified music).

I am always wondering about audiophiles who have little experience with going to concerts of unamplified live music (e.g., classical) and mostly listen to rock and pop: what does it really mean when they prefer LP over CD? I guess it says more about their tastes, which I have sympathy for, and perhaps about the relative qualities of their LP vs. CD playback systems, rather than it says anything about a reasonable comparison with a valid reference. What does it mean when they say "LP is better"?

Al. M.

Better sound reproduction is a statistical matter of preference analysis - the best essay on it I know about can be found in the F. Toole book "Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms".

Or just as said by Nelson Pass "We want our products to invite you to listen. We want you to enjoy the experience so much that you go through your entire record
collection - again and again. This, by the way, is a very strong indicator. " Some old articles by Harry Pearson in TheAbsoluteSound about these matters are also very interesting.

BTW, what is your LP system at home?

 

[microstrip]

One has to be careful not to confuse 24 bit fixed-point PCM and 32-bit floating point. 32-bit floating point actually has less precision than 24-bit fixed point.

Julf,
Isn't the 32-bit floating point used in audio workstations just the IEEE 754 single-precision binary floating-point format?

 

[JackD201]

Yes, as the article says that I quoted in my OP:

While professional editing, mixing, processing, equalizing and level shifting usually use more data bits for computation (24 bits linear, 32-bit floating point or now 48-bit linear), 16 bits is more than enough for unlimited fidelity as a release format.

The reason we use more bits in production is so we can create and preserve a true 16 bits through the whole process after all the truncation and rounding and nastier stuff that goes on between the microphone and your CD.

"more than enough for unlimited fidelity"

Wow. Is that intentional or just bad english?

 

[cooljazz]

Yes, as the article says that I quoted in my OP:

While professional editing, mixing, processing, equalizing and level shifting usually use more data bits for computation (24 bits linear, 32-bit floating point or now 48-bit linear), 16 bits is more than enough for unlimited fidelity as a release format.

Unlimited fidelity in 16 bit is great! But what's better yet, is even less yet limits with 96/24 and even more with 192/24!!

What's better listening than more than unlimited...?

CJ

 

[JackD201]

I know right? Avid just released its new 64bit platform so they can do for 24 what 32 was doing for 16. Preserve it.

 

[Julf]

Isn't the 32-bit floating point used in audio workstations just the IEEE 754 single-precision binary floating-point format?

Yes. So it is the smallest-size widely-accepted floating point format. They use 32 bits because anything less is harder and slower to do.

 

[Julf]

Yes, as the article says that I quoted in my OP

Yes, I was actually agreeing with you

 

[Julf]

I know right? Avid just released its new 64bit platform so they can do for 24 what 32 was doing for 16. Preserve it.

No, it seems that the reason they went for 64 bits is that 64 bit arithmetic is actually faster than 32 bits on modern processors.

 

[microstrip]

"more than enough for unlimited fidelity"

Wow. Is that intentional or just bad english?

Jack,

You can always search yourself for the expertise and credibility of the author of these (and other) words. I did it since post 1.
http://www.kenrockwell.com/

 

[microstrip]

Yes. So it is the smallest-size widely-accepted floating point format. They use 32 bits because anything less is harder and slower to do.

Julf,
If so I fail to understand why you say it is has precision than 24 bit PCM - a 32 bit floating notation can represent values of less than 1 with enormous precision, down to 1.175 x 10-38 . It has 23 bit mantissa, but also an 8 bit exponent.

 

[Julf]

Julf, If so I fail to understand why you say it is has precision than 24 bit PCM - a 32 bit floating notation can represent values of less than 1 with enormous precision, down to 1.175 x 10-38 . It has 23 bit mantissa, but also an 8 bit exponent.

Yes, I have to correct myself. 32-bit IEEE 754 floating point doesn't have less precision than 24 bit linear fixed point - but it also doesn't have any more precision.

The *precision* of a 32-bit floating point is 24 bits. The *range of values* it can represent is larger than 24 bit linear fixed point.

 

[JackD201]

No, it seems that the reason they went for 64 bits is that 64 bit arithmetic is actually faster than 32 bits on modern processors.

ergo.............