I listened to the link you provided, but I'm sorry to say it didn't provide any answers. 300Hz is very low frequency and therefore very responsive to dither.
Dithering is merely changing the LSB to a form of single-bit or bitstream conversion - it's not magic, and as with all single-bit conversions it is severely bandwidth limited at the available rate of 44.1kHz.
So yes, for low and lower-mid dither gives you extra resolution, but in the treble it's simply noise.
For me this was borne out in the music samples too - treble was terrible in all 3. Good article though, thanks for the link. As I said before: there is a reason SACD and 24/96 digital sounds better! Records are of course subject to a huge range of quality issues all the way through the chain: however it's probably about time that people acknowledged that CDs are too, but the CDs failing are rather less excusable in 2010
abraxalito: I think we are getting too aggressive with each other so I'll not be replying to any more of your posts, I'm sure you'll understand.
My experience is exactly the opposite.... top notch digital beats mediocre Analog .. only ...
+ kabizllion ...........
This is exactly what i hear with most CD's , not all ... The really good ones have good dynamics but still nothing to Analog. Everytime I'm fooled into believing that digital have arrived and then give analog a listen, then bam it hits you , more dynamics, much more realistic in it's reproduction than digital.
Unfortunately the convenience and repeatability of digital still keeps it out front as the compromise to have ..
regards,
Pages & pages of discussion about the number of bits. The problem with CDs is not bits, but sampling rate. 16 bits is sufficient in most cases (ear/brain starts running into a limit somewhere between 20-22bits).
Time resolution is more critical. Both from an ability to detect it aurally and from all that we have to put up with because the sampling on CDs is so low.
Mathematically the algorithms used require absolutely no input signal above the Nyquist or things don't work out like they should. Almost all recording sources will have signal in excess of 22k. Filters to remedy cause all sorts of other problems.
When i read the 1st Sony CD white paper in 1980 i happened to be taking a grad level digital sampling course. My immediate response was that they were going to have to quadruple the sampling (at a minimum) to approach analog. As we get higher rez material this seems to be being borne out.
dave
Time resolution is more critical. Both from an ability to detect it aurally and from all that we have to put up with because the sampling on CDs is so low.
Mathematically the algorithms used require absolutely no input signal above the Nyquist or things don't work out like they should. Almost all recording sources will have signal in excess of 22k. Filters to remedy cause all sorts of other problems.
When i read the 1st Sony CD white paper in 1980 i happened to be taking a grad level digital sampling course. My immediate response was that they were going to have to quadruple the sampling (at a minimum) to approach analog. As we get higher rez material this seems to be being borne out.
dave
No. SY was referring to one I wrote for this forum. But it's been stuck in the "Coming Articles" queue like, forever. Might see it by Christmas.
No, it's not.
Nope. It works the same at any frequency below the Nyquist frequency. As it happens, the Wikipedia article on dither is actually pretty good- you might want to start by reading that, then if you need to get into more detail, you've at least got a basis: understanding what dither is, which you don't yet.
No aggression intended towards you, but I do get that you might have interpreted my words as such - the comment about sums not being your strong suit was a tease, not an attack. I understand that you don't have answers to my points, yep
As SY has said, do some reading up on dither as you haven't grasped it yet. Also in regards to your points about dither not working at higher frequencies, ask yourself why you'd wish limit its operation to a much shorter timescale (a 1000X reduction) than you allow at LF.
I think there are just a few superb vinyl LP's. I own a few. I think there are a few superb CD's- I own a few. One can't blame the "compression" wars on the CD format, that is an artifact that most music was bought for car systems in the 00's- a bad place to listen to uncompressed music.
I've bought about 150 used Lp's in the last year. One was superb, and one was pretty good but showed it's 1953 engineering and microphones. The superb one cost me $.05, as it was an "easy listening" LP, Peter Nero on RCA Dynagroove.
Magazines always told us that the LP was limited to 55 db s/n. I think the superb albums exceeded that a little, but they were always few and far between. The only LP plants that put out really quiet surfaces, I bought were RCA, Colombia, Mobil Fidelity (don't know whose plant), and Telefunken. In order of bad surfaces when new from worst to okay, Parliament, Capitol, ATCO, Warner, Angel, MGM.
I have read that there is only one US LP plant still operating, in Nashville. I have quite a few bluegrass albums from local bands from the late 90's that used that plant, I am sure. The surfaces were okay, but not superb. I bought "Abbey Road" on new vinyl last year. The surface was much improved over Capitol product, but just okay.
I just listened to "The court of the crimson king" mentioned earlier as a CD with loud and quiet parts. While it has a strange sound, I think that is more the board equalization and microphones than the CD quantization process. It is certainly level shifted, as the solo flutes and harpsichord are louder than the drums, for example.
I find the denegration of CD to be funny when 95% of the world has dived into IPOD and MP3, a process that, so far to me, sounds about as bad as audio cassettes.
For A/B comparison I have lightly used vinyl of Serkin 3 Beethoven Sonatas and E Power Biggs Toccata and Fugues, on Colombia, one of the best of the LP labels. By lightly used, I mean I never played them above 1.5 grams, and always put them away to keep the dust off. They sound about the same to me.
I have Polovetsian Dances played by the 68 Texas Music Educators Orchestra on Silver Reed records, and a Telearc CD. The CD sounds way better. The LP was much compressed to fit it on the vinyl, I know because I was onstage listening when it was recorded.
I've owned an AR turntable with AT, later Grado FTE cartridges. Now I own a BIC turntable I bought for $69 in 1979, with Shure V15 Era IV cartridge. The latter combination is the best I've ever heard, including all my friend's systems. It plays stuff the earlier setup would not pretend to play, for example certain high velocity ATCO LP's. I own a RCA CD player purchased in 1980, and although I have bought others (Memorex, Sony), I find the RCA player just fine. This was before the 1 bit converter era. I listen through a PAS2 tube preamp, and ST120 or CS800S power amps. The ST120 is modified with modern Ft output and driver transistors, also the Djoffe bias current mod. I also have a mixer a RA88a that I use with better volume control than the PAS2, but it still produces a little more hum, so it is not my reference system. The best speakers I've ever owned are my current SP2-Xt's. My hearing is limited to 14Khz, like most adults, so what happens at 20 khz is obviously no longer important to me. My hearing is good enough to tell that the best LP's and CD's are better than any FM tuner I have ever owned, including a Dyna FM3. They are also better than HD TV over its limited built in sound system.
Last edited:
Mostly agree with you except for one significant error. There were no CD players in 1980. I saw a prototype demo at summer CES in Chicago in '82, bought my first digital recorder (Technics SVP100 14 bit EIAJ with built in VHS transport) in September 1982 and my first CD player Sony CDP-101 on Friday April 1, 1983 for only $750. I remember it well as it was April fools day and I picked it up 10 minutes before store closing time and took it home with no disc to try. Didn't get any discs until Saturday morning - and they were $18.50 a pop in 1983 dollars!
Biggs was a good organist but LP organ recordings are absolute crap compared to CDs. It's a pity he didn't live into the digital age. The other thing that LPs don't do worth a dime is transients like orchestral bells and acoustic guitar. My roommate back in the '80s went to answer the phone when I was playing Glenn Millers 'In the Digital Mood' track Pennsylvania 6-5000 and the 'phone' rang on the recording. As he picked up the phone I just smiled. I was NEVER at any time fooled by LPs but I have been by good CD recordings. Pianos suck on LPs as well.
What is the crazy fascination with the LP? Seems like more of the emperors new clothes.
I also don't miss drum brakes
G²
Biggs was a good organist but LP organ recordings are absolute crap compared to CDs. It's a pity he didn't live into the digital age. The other thing that LPs don't do worth a dime is transients like orchestral bells and acoustic guitar. My roommate back in the '80s went to answer the phone when I was playing Glenn Millers 'In the Digital Mood' track Pennsylvania 6-5000 and the 'phone' rang on the recording. As he picked up the phone I just smiled. I was NEVER at any time fooled by LPs but I have been by good CD recordings. Pianos suck on LPs as well.
What is the crazy fascination with the LP? Seems like more of the emperors new clothes.
I also don't miss drum brakes
G²
There is a Big difference between players made before 94 and what is currently available. I would suggest going with something more current for a digital reference .....
True 44 K is a problem and this is very much known through out the industry, so why haven't they ( the industry) made strides to make 96 more available ? why 24 over 16 bit, advantage ?
I would disagree with you on that , very much so and i think you would 2 , if you heard it .....
Wow! Maybe it's time you listened to a decent vinyl rig.
It's called music.
Neither do I.
I haven't read the whole pdf (yet) but this seems a decent guide to dither, complete with sound files.
http://www.izotope.com/products/audio/ozone/guides.html
/Hugo
http://www.izotope.com/products/audio/ozone/guides.html
/Hugo
So what you are saying is that converting 9 bits to 1 bit using random noise is different to converting 16 or 24 bits to 1 bit using random noise ?
To me it looks like exactly the same mechanism TBH, except for the number of bits you start with. Maybe it's just me
. Additionally I notice a common theme: all the examples ever used are for very low frequencies. 100Hz, 300Hz etc.Can you find any examples of 6kHz and above? I can't! I'd be interested in hearing them if you can though..
Which brings me to Planet10's point which I agree with: the very low sampling rate on a CD is a much bigger issue than bit resolution, in fact 16bit with dither at 192kHz would probably work well over the entire audio band.
Both dither and bitstream conversion are the shifting of information from level information into the time-band (spreading it out into many samples), something not possible to do with high frequencies at low sampling rates.
Personally at home I upsample the 44.1kHz to 88.2kHz using an UltraMatch so the DAC at end has a chance. This means I'm listening to what Behringer's idea of where the 'missing' samples would fit, but it still sounds better to me than any CD player I've heard, although I'm sure some of the better ones do this too. It still does not get rid of the horrendous clipping of course (which, ironically, is very audible on this gear) so I still need to run at reduced levels or a declipped waveform.
My records still have more dynamics than the 96dB CD system of course
. Also with records I know there is more information in there than I can extract with today's technology. Almost the opposite is true with CDs.
Regardless of any technical explanations I have yet to meet a sound engineer who would not agree that going from 16 to 24 bits yields a far greater improvement in SQ than going from 44.1/48 kHz to 88.2/96. Of course both would be preferable.
George Massenburgs digital plug-ins run at 48bit/96k because of this and when George talks I tend to listen. He has forgotten more about these things then (almost) any of us will ever know as a sound engineer, producer, lecturer, inventor (parametric eq) and manufacturer (both hard- and software).
George Massenburgs digital plug-ins run at 48bit/96k because of this and when George talks I tend to listen. He has forgotten more about these things then (almost) any of us will ever know as a sound engineer, producer, lecturer, inventor (parametric eq) and manufacturer (both hard- and software).
I hope that they'll cover dither next so that you don't fall into the same mind-set errors as some of the posters in this thread. You're still young enough to grasp the concept quickly and make it part of your intuition. Hugo's links were clear, interesting, and quite fun to go through, so you can get a chapter ahead of the class painlessly.
Same here, but... that may well indicate that the disc was mixed/mastered with less compression. The whole point of compression is to make things sound LOUDER like most people want.
edit: OMG, this is great!
http://www.izotope.com/products/audio/vinyl/
- Status
- Not open for further replies.