I first worked with PCM audio in 1970 for US Army telephone systems (mountaintop to mountaintop). No dither and nobody'd ever heard of it.
When CD first came out, we had the Kyocera, the first generation Philips-based players, the Yamaha CD-X1 (I think...) at the retail store where I was, and I bitched as much as anybody about the sound.
But that was then, this is now. If folks only complain but don't do anything nothing improves.
WRT John's scepticism, let me just say that the mathematical result of an ideal bandlimited, properly dithered A/D/A conversion is a exact replica (in all respects, at all levels including below the smallest bit, and at all frequencies - exact) of the original, plus a small noise, plus a time delay. Period. That's the ideal.
To make progress in this field we need to keep in mind that observed imperfections are not fundamental to the A/D/A conversion, but are instead artifacts of a particular implementation.
Maybe it will take a Blowtorch quality implementation to make the most discerning happy, but that shouldn't be much of a surprise considering the difficulty of the task.
Thanks, as always,
Chris
When CD first came out, we had the Kyocera, the first generation Philips-based players, the Yamaha CD-X1 (I think...) at the retail store where I was, and I bitched as much as anybody about the sound.
But that was then, this is now. If folks only complain but don't do anything nothing improves.
WRT John's scepticism, let me just say that the mathematical result of an ideal bandlimited, properly dithered A/D/A conversion is a exact replica (in all respects, at all levels including below the smallest bit, and at all frequencies - exact) of the original, plus a small noise, plus a time delay. Period. That's the ideal.
To make progress in this field we need to keep in mind that observed imperfections are not fundamental to the A/D/A conversion, but are instead artifacts of a particular implementation.
Maybe it will take a Blowtorch quality implementation to make the most discerning happy, but that shouldn't be much of a surprise considering the difficulty of the task.
Thanks, as always,
Chris
Last edited:
Yes, but some A/D/A conversions are easier to implement of higher quality than others. Like, using some components and topologies is easier to implement higher quality power or mic amps.
? The text you quoted gave 5-10 db for the headroom and gave levels higher than I measured.
Having done live symphony reinforcement is where my preferred 20-30 db comes from. The text mentions comparing the levels on an oscilloscope. That most likely would have been done with a recording.
Again we are basically in agreement! You can hear signal below the noise level in both real and digital systems. Where we differ is in what the noise level is. When you measure acoustic noise there should be an "A" weight filter in use. The noise from a DSP system is not specified that way. Good practice would be to record the signal with the record system noise floor below the signal noise floor.
The text you showed reported 104 in the 7th row of the balcony in an NC20 space. That would put the front row about 125 peak with a noise floor of 15 db above 3K.
Last edited:
The 105dB SPL of an audience seat for an orchestral performance is stated as an "instantaneous peak". For recording, headroom above this is of course required, for all the usual reasons. But we're discussing storage, where no (or 1/2 dB or so) headroom is required.
(Nobody would suggest recording at 16 bits these days, although it can be done perfectly well and with an effectively electronic noise free background when making live recordings. Audiences are loud too. But that's a completely separate discussion.)
Thanks,
Chris
That requires a rather sweeping generalization from free air conditions to concert hall conditions. Not at all certain that I can buy it.
Thanks,
Chris
Chris
The text gave 104A SLM reading! That's loud. We agree recording needs to be done with more than 16 bits.
Where we disagree is that there actually are some situations where more is required for reproduction. Of course since virtually all the recording gear is advertised as 24 bit, it would be nice to have the storage method to match
Just for grins a Shute Beta58 microphone produces 2.6 mV at 94 db and is rated at 150 ohms. What is the S/N from a recording studio noise floor to a heavy metal singer with lips on the windscreen?
The text gave 104A SLM reading! That's loud. We agree recording needs to be done with more than 16 bits.
Where we disagree is that there actually are some situations where more is required for reproduction. Of course since virtually all the recording gear is advertised as 24 bit, it would be nice to have the storage method to match
Just for grins a Shute Beta58 microphone produces 2.6 mV at 94 db and is rated at 150 ohms. What is the S/N from a recording studio noise floor to a heavy metal singer with lips on the windscreen?
The hall is a well known space is where the NC numbers come from. The curve loss is from memory.
The 10db path gain is from critical distance and the other 10 is SLM-peak.
Then there is the story of the concert hall that put in two 4" conduits to allow for recording. It destroyed the NC rating!
Last edited:
When they get through "mastering" it? Maybe 10dB - it'd be funny if it weren't so true these days. I can't disagree that a "24 bit" format would be plenty good enough for everybody, and the CD sampling rate is really not wonderful, but I'm worried that the attention of folks in a position to move things along could easily be distracted from the terrible state of so much professional mastering these days while chasing tiny marginalia.
Listen to any of the 2008 Van Morrison reissues and tell me that even 8 bits is needed anymore. Or recent EmmyLou Harris remasters, or lots of others. Sinful.
Thanks, as always,
Chris
And that's so sad ...
I first got the good stuff, satisfying sound from digital, CD, over 25 years ago, and knew from then on that it was all about the implementation, as others here have stated. At times I have been staggered by how gratingly and tediously unpleasant very expensive digital playback can sound, and there's no reason for that; apart from the fact that many see top notch measurements routinely achieved and claim that therefore all is solved, nothing more needs or can be done. Well, there is just a bit more to it than that: digital is a hard taskmaster, and if you don't treat her right, she'll bite you in the bum. Or bore you to death ...
Frank
Last edited:
Do tell. Exactly what digital set up provided you with satisfying CD playback over 25 years ago? Which would have been around 1987, only about five years after the introduction of the format.
This too shall pass.
Whether it will do so while we are still alive is an open question.
One of the earliest digital recordings was the "Cleveland Winds" doing Holst and Handel and Bach, under the baton of Freddie Fennell, with Stockham's 50k 16 bit sampling box. But I suspect he and his folks really labored over the hardware to ensure monotonicity. Even with the sample rate conversions the stuff sounds o.k. to me today, if not the greatest ever. But then I do love the Holst, so I'm inclined to forgive.
Last edited:
As I've made clear many times, it's all about worrying about the small stuff that makes it happen. And simplifying. The system was a Yamaha CDX1100, the top of line unit for the day with digital volume control, directly driving a Perreaux power amp, and B&W bookshelf speaker. This was all hard wired, from the power feed in the room wall socket right the way through to the speaker drivers. A ritual of constantly finding and eliminating little weaknesses was absolutely essential, and still only guaranteed top notch performance for a relatively short period of time, the sound quality would slowly degrade following a process of effectively resetting the electronics.
But it was enough to make me aware of what could be achieved. Back then I knew far less than now, so the situation now is a lot more in hand ...
Frank
Last edited:
There is none. The bandlimiting itself can cause variable group delay issues, depending on implementation, but it's not inherent in the A/D/A conversion process.
I know I must sound like a broken record, but we need to keep focussed on the real issues. And yes, I do still have over fifty feet of vinyl records and a good player. But I certainly wouldn't choose to buy a new vinyl record now rather than a CD if both were correctly mastered.
Funny, nobody ever complains about the group delay issues of vinyl records, although they're well understood.
Thanks,
Chris
So you have an ideal brickwall AAF with no overshoot in the time domain whatsoever?
- Status
- Not open for further replies.