You conveniently skipped my previous question, so I have to ask again: what exactly do you mean by "hearing" in this context?
Were you able to compare the same source material in 16 bit dithered vs. 16 bit non dithered in a DBX test and you were able to sort out the sources to a statistically significant level? Not a big deal to set up, and this one I may believe you can...
However, the discussion was about audio dynamic range, nothing to do with dithered/not dithered audio formats. It was shown that nobody (so far) was able to identify in a DBX test a -80dB (14 bit) stray signal in a musical program, this is what the Sousa band test is all about.
Completely different DBX tests, completely different results relevance, apples and pears, if you prefer.
These are two different points. Lack of dither is the same as (actually worse than) bad crossover distortion. At 20dB below full scale we more at 12bits. This is not 16bit level aberrations in the presence of 0dB content.
Waly - It is not that hard with some program material to hear lack of dither at 44.1/16.
Last edited:
For want of any better way to explain it, a lack of dither to me makes the program material sound a little rough-edged, or perhaps grainy. Problem with this the type of explanation I can give is that it probably won't mean anything to you unless you have heard it yourself. Same way you wouldn't know what yellow looked like if you had never seen it before.
Just to dig myself in a little deeper, maybe I am a glutton for punishment, suppose you have just shaved in the morning and you run your hand over your chin in the morning to make sure. It feels appropriately smooth, so no stubble left. But, I don't feel for stubble the same way you do. I check for it more carefully in different directions and pulling this way and that while rubbing, and lo and behold I noticed a little stubble you might never notice the way you do it. In fact, I have trained myself to do a very thorough stubble examination, so much so that I do it without even thinking about it. I can hardly not do it, it's just automatic. But, once I feel it the exact right way it feels pretty rough. And so, maybe I say you would have to have numb fingers not to notice it, because I can't imagine you would do it any other way than the way I do, and so I expect you to notice it the same way I do. (end of metaphor).
Why do I even notice this stuff? Long time ago I had a job as a sound man. I was very competitive at that age, and other people in the business claimed to hear little things I didn't notice. Not to be bettered, I practiced until I could hear if the little aberrations were there or not. Other people who were in the same line of work and I sometimes talked about things we noticed or didn't notice, and we found that we independently heard the same things or didn't hear them, under the similar circumstances, such as maybe the little quirks of a particular compressor that most people wouldn't ever notice. But, we found we our impressions from having listened to the compressor at different times and places from each other was that we noticed the same little things, and, again, did so independently. The point is, we noticed things independently and only later found out we agreed on what it sounded like.
Lots of other things have happened over a lifetime to give some confidence that I am not imagining things. People still send recordings they made or are working on for me to check. They tell me I notice little things they don't. Once I point little things out, or in the worse case, coach and tutor somebody how to listen then they usually hear it too. They then find they can fix it with digital editing tools, once it has been noticed. Then it's gone for both of us.
Lots and lots of stuff happened and continues to happen to convince me, but not you, that I am mostly not crazy. And, no problem saying I do make mistakes sometimes, too. It happens. Absolutely. Try to learn from them when they do.
Okay, time to wind this down. I didn't want to make this about me bragging about my hearing or my playback system. That would be a total waste of time. I'm sorry to have digressed this much. I've told you about as much as I am going to about hearing dither and other things, and I would like to end it here.
Let's find somebody else younger and with better hearing to use as a test subject to prove this or that about what some humans may be able to hear. I'm not interested, and would be a lousy choice anyway.
Sure, of course it would have be some explanation that makes sense. It is relatively easy to hear. So is some other stuff that some people claim to hear, although most of those things, or perhaps all of them, are not things that some trusted source probably can verify. Much of it has not been seriously studied, and certainly not to my knowledge at all using modern state of the art audio gear as test equipment.
Since we don't have a mutually trusted source to confirm anything beyond dither in 16-bit systems, seems like we will continue to have different views on what may be audible for some people.
Despite that we might disagree on a few things, I might be inclined to agree that hearing Sousa bands at -60dB seems unlikely. Haven't tried it, but it seems like it might not be possible.
Possible, but I would love to put that one to bed. If you ever are able to measure FR at your listening position it would be of interest at least to put to bed one of the things thrown at NS-10 performance.
Scott,
As usual. I mentioned a sample rate that as far as I know would not cause any perception issues in that part of the chain. I don't think anyone with experience in the field would argue that. The issue is basically how much you can go below that and still not give away the illusion.
There are so many other variables in a reproduction system that make pretty much all testing limited to specific comparisons.
The school you seem to oppose are the "This sounds better" crowd. They don't seem to isolate each variable. Quite simply I don't see the point in arguing what sounds better to any given individual.
My concerns are what I have to do to get the best reinforcement or reproduction out of what is available to me.
Making my house quiet was just attention to details. Most folks don't know a simple seam along a window unit without a proper sealant will let in more noise than the entire rest of everything in the wall.
You are aware I aim for -160 dB on electronics artifacts. Doesn't mean I will get there on a system, but it doesn't hurt.
I have the equipment to measure most of what I do. Now many such measurements can be made to the point of absurdity. But other measurements that are not done may have more influence on perception at smaller levels. The issue is to remove the noise of the complaints to get to the signal of what is actually distorting perception.
Now as to Fletcher Munson weighting, musical spectra and TEF waterfalls they are well known and often exploited to advantage. Where do you think analog equalization standards came from?
Now you probably have never seen one of the older reinforcement specifications that called for a 3dB per octave roll off above typically 3,000 hertz. Used to be common. There even were reasons why it was so. And those folks who understood the reasons knew when not to follow it much to the concern of folks who specified it, because that is the way things were done.
Yes the records I have sound quite good to me on my gear. I also have CDs that clearly were made from older records, the music is what I want and it is only wishful thinking that I want better quality. Now I also have some audiophile CDs that to me sound horrible. Of course I also have CDs that sound great.
Now one if the interesting quirks of what I do is the takeaway others give me. There was the stadium where a group of top notch studio recording guys all were there to equalize it. They all came back with different corrections. Seems they didn't pick up they were listening to hundreds of different loudspeakers depending on where they stood. Final change 1/2 dB on a single 1/3 octave filter.
Different folks hear different things!
As usual. I mentioned a sample rate that as far as I know would not cause any perception issues in that part of the chain. I don't think anyone with experience in the field would argue that. The issue is basically how much you can go below that and still not give away the illusion.
There are so many other variables in a reproduction system that make pretty much all testing limited to specific comparisons.
The school you seem to oppose are the "This sounds better" crowd. They don't seem to isolate each variable. Quite simply I don't see the point in arguing what sounds better to any given individual.
My concerns are what I have to do to get the best reinforcement or reproduction out of what is available to me.
Making my house quiet was just attention to details. Most folks don't know a simple seam along a window unit without a proper sealant will let in more noise than the entire rest of everything in the wall.
You are aware I aim for -160 dB on electronics artifacts. Doesn't mean I will get there on a system, but it doesn't hurt.
I have the equipment to measure most of what I do. Now many such measurements can be made to the point of absurdity. But other measurements that are not done may have more influence on perception at smaller levels. The issue is to remove the noise of the complaints to get to the signal of what is actually distorting perception.
Now as to Fletcher Munson weighting, musical spectra and TEF waterfalls they are well known and often exploited to advantage. Where do you think analog equalization standards came from?
Now you probably have never seen one of the older reinforcement specifications that called for a 3dB per octave roll off above typically 3,000 hertz. Used to be common. There even were reasons why it was so. And those folks who understood the reasons knew when not to follow it much to the concern of folks who specified it, because that is the way things were done.
Yes the records I have sound quite good to me on my gear. I also have CDs that clearly were made from older records, the music is what I want and it is only wishful thinking that I want better quality. Now I also have some audiophile CDs that to me sound horrible. Of course I also have CDs that sound great.
Now one if the interesting quirks of what I do is the takeaway others give me. There was the stadium where a group of top notch studio recording guys all were there to equalize it. They all came back with different corrections. Seems they didn't pick up they were listening to hundreds of different loudspeakers depending on where they stood. Final change 1/2 dB on a single 1/3 octave filter.
Different folks hear different things!
You know since I was played this I never tried to find it and my friend and his wife both have passed away very young to cancer. He also played me this Horowitz channeling four hands YouTube You can't destroy great music, I think this was an AM radio broadcast recorded on a wire recorder.
Darn it! Again, people dying and taking info with them. My dad did this to me also. This needs to stopped!
Thanks George that was lovely. Is it just expectation bias, knowing it was 1937, that makes the choral parts sound a bit creepy to me? Like watching the picnic scene in Cabaret.
Last edited:
Ok, so tell me if my understanding of your ueber verbose answer is correct.
Imagine the following gedankenexperiment: I take say 24 random pieces of music of your choice, that were originally recorded say 24bit/196KHz, and I truncate them all to 16bit. Then I take randomly half of the 16bit files files (12) and dither them by a random method (say RPDF, TPDF, Gaussian PDF, Noise Shaping). Then I hand over to you the 24 files.
You claim that you will be able to sort them out, 12 files undithered and 12 files dithered, with less than (say) 5%...10% error?
Of course, I'm not expecting you to correctly identify the dithering method.
Last edited:
As I answered before I haven't done this. But Steve Hoffman did:
Quoted from Here
" First, let me say that I love records, compact discs and SACDs; I have a bunch of all three formats. Nothing that I discovered below changed that one bit.
I did these comparisons a few years ago. Since I spilled the beans to an interviewer on mic last year I continually get quoted and misquoted about this subject. I'll try to set the "record" straight in this thread. Please note I'm typing on a whacked out computer not my own with a tiny monitor and no spell check.... There could be a (gasp) typo or two...
A few years ago, mainly out of curiosity (and nothing else) I got the chance at AcousTech Mastering to compare an actual master tape to the playback of a record lacquer and digital playback. Also did the same test using DSD (SACD) playback as well later on in the day. The results were interesting. The below is just my opinion. Note that we cut the record at 45 because the lathe was set for that speed. A similar test we did using the 33 1/3 speed yielded the same result.
FIRST COMPARISON: MASTER TAPE with ACETATE LACQUER AT 45 RPM with DIGITAL PACIFIC MICROSONICS CAPTURE.
We had the master tape of the Riverside stereo LP Bill Evans Trio/WALTZ FOR DEBBY at AcousTech and decided to do this little comparison. Since the actual master needs a bunch of "mastering" to make it sound the best, I set the title track up as if it was going to be mastered (which in a sense it was, being cut on to an acetate record).
We cut a lacquer ref of the tune with mastering moves while dumping to the digital computer at the same time with the same moves.
Then, after a break, we sync'd up all three, first matching levels. Simultaneous playback of all three commenced and as Kevin switched, I listened. (We took turns switching and listening). First thing I noticed:
The MASTER TAPE and the RECORD sounded the practically the same. We honestly couldn't tell one from the other during playback. This was of course playing back the tape on the master recorder with the mastering "moves" turned on. The acetate record was played back flat on the AcousTech lathe with the SAE arm and Shure V15 through the Neumann playback preamp (as seen in so many pictures posted here of AcousTech).
The flat digital playback of my mastering sounded different. NOT BAD, just different. The decay on the piano was different, the plucks of Scott's bass were different, the reverb trail was noticeably truncated due to a loss of resolution. Non unpleasant, just not like the actual master tape. This is slightly frustrating to me because it confirmed the fact that when mastering in digital one has to compensate for the change (which I do with my usual "tricks"). The record however, gave back exactly what we put in to it. Exactly.
Please note that an actual record for sale would have gone through the manufacturing process and the lacquer would have been processed to a MASTER, MOTHER, STAMPER and VINYL with increased surface noise, etc. but the sound of the music remains intact for the most part. A remarkable thing since records have been basically made the same way for over 100 years.
SECOND COMPARISON: MASTER TAPE with ACETATE LACQUER AT 45 RPM with DSD MASTER (SACD MASTER).
So, using the same master tape of WALTZ FOR DEBBY, we compared the before mentioned acetate that we cut on the AcousTech lathe (manufactured in 1967 and modded by Kevin Gray) with a DSD playback of the same tape with the same mastering and levels.
Result? The DSD/SACD version sounded even MORE different than the compact disc digital playback compared to the analog master. More not-like the sound of the actual master tape. The resolution was fine and we could hear the notes decay, etc. just like analog but the TONALITY was a bit off. It was not telling the truth when compared to the master tape or the acetate record.
THIRD COMPARISON: MASTER TAPE with ACETATE RECORD with OPEN REEL TAPE COPY AT 15 ips:
We made a dub of the tune WALTZ FOR DEBBY to an Ampex ATR-100 at 15 ips non-Dolby, +3 level and played it back with the actual master tape and the acetate record. Both of us thought the open reel tape copy sounded inferior to the acetate record when compared to the master tape; weaker transients, a more "blurred" sound that would never be noticeable unless played back with the actual master tape to compare it to.
So, what does this mean to you? Probably nothing. What did it mean to me? I found it interesting. The CD playback had more accurate tonality than the DSD/SACD playback. The DSD playback had more front to back resolution than the CD playback. The tape copy sounded slightly lackluster. The acetate record playback beat them all in terms of resolution, tonal accuracy and everything else when compared directly with the analog master in playback. This is not wonderful news in a certain sense; vinyl playback is sometimes a pain in the butt and knowing that CD's are not capturing everything in perfect resolution drives me bonkers.
Regarding the lowly phonograph record:
We know that records have their problems (could be noisy, warped, bad cutting, etc.) as well but for the most part they will be a damn miraculous representation of the actual master recording for not much money.
Your comments are welcome.
Please remember, the above is just my OPINION but I found it interesting. I love my compact discs but I realize they are not the last word in resolution; they are damn fine though and when listening for pleasure I play CDs and records, with CDs getting the most play. My Sony and Living Stereo SACDs are never far away from me either. If you disagree with me, that's cool. It's all fun, or should be."
Last edited:
"Truncating" is eliminating the least significant N out of M bits (here, 8 out of 24). You cut those bits, then you dither (or not) the result. Essentially, dithering replaces the new low levels with some sort of noise that could be perceptually less annoying than the LSB toggling 0 or 1.
So can you? Careful with your answer, somebody may hold you up to it
.Maybe there is better today. It was and still is a Pacific Microsonics Model 2 A/D D/A converter. Here is Kevin's gear list
Dither first, then truncate. "Dither should be added to any low-amplitude or highly periodic signal before any quantization or re-quantization process..."
Dither - Wikipedia
But, to answer your question, if properly dithered and truncated, or only truncated without dither, starting from clean Hi-Res sources, I have heard it so far and I expect I would again. Don't know why not. Is there some trick I don't know about?
Thanks Morinix for your INFORMED LISTENING OPINION based on some of the best audio playback equipment and an OPEN MIND! This is how it is done folks! Not some double-blind lab with people paranoid that somebody somehow might get a listening advantage. Just using our ears with careful comparisons. That is how I listen, you should too.
- Status
- Not open for further replies.