The only place it has a prayer (in presentation) is passages where there is extremely wide dynamic range in the recording that is conserved rather than compressed for reproduction. But with that being 96.3 dB (assuming perfect linearity over 16 bits of course) of DR, that's asking the recorded space to be awfully quiet, too. Not sure how many microphones have that kind of DR themselves (Rode NT1's, Senn MKH030/080's).
Oh and Ed Simon's living room for listening as to not mask the quietest-est-est passages.
But who's remotely making music like that?
When I have recorded and mixed at 24-bits, and then have to make a CD from it, I always have to two files right there to compare. There is always loss of detail and increased "grainyness" in the 16-bit file. It always seems a pity to have to send it out that way. Other people there me with hear it too, and feel the same way. You can think we are all nuts or delusional if you want. As I said, there is no way I can satisfy want you seem to want with words. You seem to want to hear it for yourself. Just come over to my place and I will be happy to show you. Or, maybe there is someplace more nearby where you are like a mastering room you could visit. Otherwise, I don't think you will ever feel satisfied.
Again, though, the effect is small but there. If you were to hear it, you would probably prefer it, although sometimes too much detail can be a bad thing. It depends on the source material.
Last edited:
Bill, maybe I would add that if you have heard undithered 16-bit, it may sound to you like there is some rough, grainy hair sitting on top of all the sounds. It doesn't have to be quiet to hear the grainyness.
When dither is added, it smooths out a lot of the grainyness, but some of it is still there, and it replaces some detail that was there previously in the 24-bit source.
I still think the analogy to bit-reduced and dithered images is reasonably apt, about as much so as analogies can be. Undithered squares in an image are like grainyness in sound. Dither helps, but if you get close enough to the screen, you still see there are squares. And with sound, you still hear some grainyness, and some loss of previous detail.
I guess if you ever heard it, you might say the analogy is very apt, but right now it may not seem that way at all.
When dither is added, it smooths out a lot of the grainyness, but some of it is still there, and it replaces some detail that was there previously in the 24-bit source.
I still think the analogy to bit-reduced and dithered images is reasonably apt, about as much so as analogies can be. Undithered squares in an image are like grainyness in sound. Dither helps, but if you get close enough to the screen, you still see there are squares. And with sound, you still hear some grainyness, and some loss of previous detail.
I guess if you ever heard it, you might say the analogy is very apt, but right now it may not seem that way at all.
For the nth time, It's not the 'hearing', my confusion is over what is there to be lost. You appear not to be worried about whether there is a real difference as you trust your ears whilst listening with a 5dB peak at 2kHz, which is fine and I am not saying you can't hear it. I just can't think of a natural musical waveform that needs this for domestic listening.
I suffer from tinnitus so ultra-subtle details are probably below my detection threshold anyway. I also do odd things like listening to 60 year old vinyl recordings which I find eminently enjoyable.
But thank you for the invite. If it weren't for the 7000 miles between us and an impending arrival I'd take you up on that.
Edit: I really don't get the dither analogy in graphics (or anti-aliasing as they like to call it), as moving close to the screen is no different from sticking your ear next to the tweeter in the quiet passage and going 'hey I can hear grunge). dithered 16 bit noise floor is -130dB. At THx levels that puts the noise floor below 0dB. AKA not an issue. If it were a monitor it would be retinal.
I suffer from tinnitus so ultra-subtle details are probably below my detection threshold anyway. I also do odd things like listening to 60 year old vinyl recordings which I find eminently enjoyable.
But thank you for the invite. If it weren't for the 7000 miles between us and an impending arrival I'd take you up on that.
Edit: I really don't get the dither analogy in graphics (or anti-aliasing as they like to call it), as moving close to the screen is no different from sticking your ear next to the tweeter in the quiet passage and going 'hey I can hear grunge). dithered 16 bit noise floor is -130dB. At THx levels that puts the noise floor below 0dB. AKA not an issue. If it were a monitor it would be retinal.
Last edited:
And for the nth time, I'm telling you what is lost is some detail, and something unpleasant is added, grainyness. That's it.
FR peaks don't have anything to do with it, of course, no more than if you looked at a dithered image on your computer screen and the colors were a little off from what they should be. You would still see there are little squares.
Regarding the difference for domestic listening and whether or not it matters, I think it depends. Domestic listening at cell phone quality can be fine for some purposes. However, I can't stand the satellite radio in my car. I make maximum quality AAC files and put then on a USB stick for listening in the car.
One problem for domestic listening that I think may have been alluded to by someone else previously, is that equipment that makes small differences audible is not inexpensive. And price, and magazine reviews, are no guarantee of quality. I think the problem there is we need advances in technology to bring down the price of better performing equipment. Of course, it would help if there were a mass market for it. Unfortunately, many people just don't care, cell phones and satellite radio are fine for them. And maybe they are in fact fine for almost everyone, except for a few odd nitpickers. I don't know.
So, Bill, I can't make any argument to you that you need better than cell phone quality or satellite radio quality. Only you know what performance you find sufficient for your needs.
I thought the same thing for a long time until I actually did the exercise myself on a -90db sine wave. The dither adds pure white noise on any time scale and there is no THD at all left. That is the result is indistinguishable from white noise added to a perfect -90db sine wave, it's your guess where the graininess comes from. All arguments to the contrary on dither usually violate Nyquist, like it takes time for the dither to average away the THD, etc. So if the resulting -93 or so dB noise floor is not audible nothing else should be.
BTW how many folks have measured the noise at the output of their PA with the signal chain set at normal listening level with no input and worked it back to equivalent LSB's.
Last edited:
I suppose the grainyness is the sound of dither noise.
It always sounded to me like quantizing noise was more like distortion than noise, at least in that it sounds more like distortion and less like noise, as compared to what we usually tend to think of for those things. So, maybe dither is kind of similar, in that sense.
Of course, this is speculation. I don't know for a fact.
Regarding PA systems, the usually are not the best quality compared to studio equipment, they tend to be used in noisy environments, and they are usually listened to at a distance where there is HF attenuation, often with room, and other reflections. I would be leery about trying to use something like that for trying to hear small details, although there may be some that are okay for that, I don't know.
Last edited:
I don't buy the 'you have to have expensive gear to hear it' line. For most parts of the reproduction chain there are designs on here (DIY forum after all) that can be built for non-silly money and will wipe the floor with most commercial designs. There are exceptions of course, but I am happy that my built from kit amps with their single digit PPM distortion will hold their own driving my ribbons.
But we at least have got to the point where you have admitted no empirical or mathematical basis for dithered 16bit sounding 'grainier'. That was all I wanted to know. Hopefully someone will have a link I can ponder on this as I'm honestly interested if my assumption (not needed in the home) is in fact valid or not from a scientific/engineering perspective.
But we at least have got to the point where you have admitted no empirical or mathematical basis for dithered 16bit sounding 'grainier'. That was all I wanted to know. Hopefully someone will have a link I can ponder on this as I'm honestly interested if my assumption (not needed in the home) is in fact valid or not from a scientific/engineering perspective.
The spectrum of TPDF dither is pure white, there are some serious theoretical papers available (which I have posted for all to read in the past). I choose to leave out the noise shaped dithers to keep it simple. Maybe someone should make an infinite resolution Gaussian noise vs. TPDF dither noise file at the same rms power to compare.
BTW don't confuse dither in imaging where the pixels are either on or off and the dither is a spatial effect to create a psuedo-gray scale with dither as we are talking about here.
Now I'm confused. Obviously, 24-bit and dithered 16-bit are empirically and mathematically different. But, how something "sounds" is a mental experience mediated by system 1 processing. I'm not aware of any mathematical models for it. For people who hear grainyness, it is empirically evident, just as is the ability to hear language, or British accents. We don't know exactly how brains do any of those things at the moment, but they do, at least for some people. Were you expecting something different? And, I don't have any idea why one would characterize a brief summery of that nature to be an admission.
Right, I do get that.
I don't know why the experience of listening to dithered 16-bit audio is the way that it is, why the word grainy seems to fit, or why I also experience it as something like little, short, rough, jagged, hair on top of the sound. I have heard engineers refer to some kinds of distortion as hair, though, and somehow the word just seems to fit when certain sound is experienced. Of course, I also don't know why I experience red the way I do. So far as I know, we all take many mental experiences for granted, as though they are the reality and not just mental experiences evoked by reality.
You imply something different than simple additive white noise, which at the -93dB level is not audible in any normal listening situation. It's pretty easy to sit down together and agree on the audibility of nothing.
Scott,
I have found that with a live symphony orchestra 30 dB of headroom leaves no artifacts I can hear.
Now if I listen at 80 dB that would leave the noise floor 110-93=17. Narrow spectrum noise can be perceived up to about 20 dB below the noise floor.
However finding a symphony hall that quiet is not going to happen. Add to that the normal microphone and preamp noise to reach a reasonable limit.
So for a well done recording during playback in a normal environment 16 bits is fine. But worst case 22 bits would cover all cases. Of course as far as I know there are no true 24 bit converters that are fast enough to allow for smooth anti-aliasing filters.
Now for even acoustic folk music you don't have that kind of dynamic range.
The practical test for how quiet a room is, would be the unwrap the cellophane from a hard candy.
In my house to measure the noise floor I have to hold my breath.
Last edited:
Dithered 16-bit does not sound like the only thing different than the original 24-bit source is a little white noise. I don't know why. Also, my listening situation in that case is near field, as I keep saying. That by itself is probably not normal, at least not with a system capable of very detailed playback (thanks primarily to the DAC-1, I would say).
Anyway, I don't know how to reconcile this without some more research. I believe some other people hear it about the same as I hear it. We would probably need to identify such people and run some experiments with them to see what we could figure out. Until then, I don't expect much in the way of progress. However, I kind of expect we would find that some people can hear some things at lower percentages of distortion than previously believed, but maybe there will turn out to be some other explanation. I don't know.
Guys please don't start using the term "artifacts" for when something sounds wrong, as opposed to an addition. That is what audiophiles that can't even tell you the direction of electron flow say... the most non-technical people.
Now, the term "abberation" is correct say when a device screws up reproduction. For example a compressing driver may give an abberation of a cello, but its not an artifact till the VC makes a scrapping sound or the spider is loose and smacking stuff.
Now, the term "abberation" is correct say when a device screws up reproduction. For example a compressing driver may give an abberation of a cello, but its not an artifact till the VC makes a scrapping sound or the spider is loose and smacking stuff.
- Status
- Not open for further replies.