I may be wrong here but i suspect there is tiny minority of mainstream pop fans who care one bit about quality. And the record companies response has so far been logical - cater for the classical and jazz fans first with rock probably to follow.
sorry, 24 bit is a marketing scam
24 bits is pointless - higher sample rate automatically gets you more bits of audio S/N with modern noise shaped dither
thinking 24 bits is a needed improvement in consumer audio delivery is audio innumeracy
psychoacoustic weighted dithers already deliver >120 dB weighted S/N to the ear with 16/44
I'd choose dithered 16/96 all day long over 24/44 or 24/48 on technical grounds - but wouldn't expect any but tiny fractional % / high sigma tail of the population to be able to actually tell in controlled listening though
24 bits is pointless - higher sample rate automatically gets you more bits of audio S/N with modern noise shaped dither
thinking 24 bits is a needed improvement in consumer audio delivery is audio innumeracy
psychoacoustic weighted dithers already deliver >120 dB weighted S/N to the ear with 16/44
I'd choose dithered 16/96 all day long over 24/44 or 24/48 on technical grounds - but wouldn't expect any but tiny fractional % / high sigma tail of the population to be able to actually tell in controlled listening though
Last edited:
Unfortunately when it comes to product decisions, businesses are much more inclined to focus on P&L than petitions. That explains the current dilemma. Popular music formats, which are predominately consumed via media players and earbuds or over car radios are routinely compressed to death and MP3 quality. That seems to satisfy those customers and the compression and low-rez file sizes offer some business benefits.
Classical and jazz still maintain some focus on quality of reproduction because those customer groups demand and are apparently willing to pay for it, or at least tilt their purchases toward it when they find it.
Until you find a way to present tangible financial rewards that might tempt some of the producers and suppliers to experiment with quality focused popular music formats, the current state of affairs is likely to continue. The problem is only compounded by the fact that the listening habits of many of the earbud and in-car crowd is quickly running them down the path to deafness, which will only increase their demand for compression to fill in the empty spaces where even currently compressed music will fall below their hearing thresholds.
Classical and jazz still maintain some focus on quality of reproduction because those customer groups demand and are apparently willing to pay for it, or at least tilt their purchases toward it when they find it.
Until you find a way to present tangible financial rewards that might tempt some of the producers and suppliers to experiment with quality focused popular music formats, the current state of affairs is likely to continue. The problem is only compounded by the fact that the listening habits of many of the earbud and in-car crowd is quickly running them down the path to deafness, which will only increase their demand for compression to fill in the empty spaces where even currently compressed music will fall below their hearing thresholds.
i'm going to completely disagree, even with 16bit/384kHz if you use digital attenuation, you will still suffer issues well before 24/48 (of course I would prefer 24/96 or better)
the majority of consumers, not as many audiophiles but many (including myself) use digital volume control, you may trust your dithering implicitly to make up the bits in the hardware, but why do it when its not needed? the hardware is there in just about every persons computer, ipod and stereo. seems like a complete waste of resources in more ways than one and remastering a 16bit version for everything from a 24bit file is hours and money wasted too, its stupid...
when all DACs including the 0.50c ones these days are 24bit or higher and any new equipment will just have it by default, where is the motivation to continue marketing fluff? the engineers all use it to give more headroom in the mastering stages for this very reason, are you saying that the whole worlds audio engineers are the victims of an elaborate plot?
the only reason you can claim that 16bit can do 120dB is because its dithering it up to 24bits or above, yet you think its not needed? sorry mate I normally agree with you, but i'm not following you at all.
16/96 for delivery format - not subsequent processing
I wouldn't roll back 24 DAC for playback, Media sw internal wordlength
having digital volume applied to a 16 bit stream it makes sense to send the increased wordlength to 24 bit DAC, not having to redither
but to deliver the information to the consumer's digital playback device at all the most reasonable place to question current 16/44 is on the Nyquist frequency barely over estimates of human high frequency perception
boosting sample rate to 88 or 96 k is the side of the box that is more constraining than the 16 word length
anti-imaging/reconstruction filters are pointed to as possibly problematic – high sample pushes any problems way beyond extreme estimates of audibility
but it does requires some engineering, psychoacoustic knowledge to appreciate the practical "resolution" added by dither and the advantage of increased sample rate, BW to the dither noise shaping
http://www.meridian-audio.com/w_paper/Coding2.PDF[/QUOTE]
http://audio.rightmark.org/lukin/dither/ (both somewhat dated compared to todays best dither performance - so the numbers given are way conservative)
understanding dither shows that increasing sample rate pushes out both sides of the digital audio box at the same time - to the extent that 24/96 would simply be a waste of the lower 8 bit for human perceptoin
so given a choice I'd definitely push for higher sample rates before greater bit depth - probably too nuanced a position for a consumer campaign - but way better than thinking you actually got something of value when you pony up extra bucks for “hi rez” 24/44
I wouldn't roll back 24 DAC for playback, Media sw internal wordlength
having digital volume applied to a 16 bit stream it makes sense to send the increased wordlength to 24 bit DAC, not having to redither
but to deliver the information to the consumer's digital playback device at all the most reasonable place to question current 16/44 is on the Nyquist frequency barely over estimates of human high frequency perception
boosting sample rate to 88 or 96 k is the side of the box that is more constraining than the 16 word length
anti-imaging/reconstruction filters are pointed to as possibly problematic – high sample pushes any problems way beyond extreme estimates of audibility
but it does requires some engineering, psychoacoustic knowledge to appreciate the practical "resolution" added by dither and the advantage of increased sample rate, BW to the dither noise shaping
http://www.meridian-audio.com/w_paper/Coding2.PDF[/QUOTE]
http://audio.rightmark.org/lukin/dither/ (both somewhat dated compared to todays best dither performance - so the numbers given are way conservative)
understanding dither shows that increasing sample rate pushes out both sides of the digital audio box at the same time - to the extent that 24/96 would simply be a waste of the lower 8 bit for human perceptoin
so given a choice I'd definitely push for higher sample rates before greater bit depth - probably too nuanced a position for a consumer campaign - but way better than thinking you actually got something of value when you pony up extra bucks for “hi rez” 24/44
Last edited:
Very well.
So we are not only to stop the use of compression in all music, we are also going to make sure recording studios re-write the sw to support dithering properly.
There is a difference in sound quality when rendering music in 24bit/48kHz and 16bit/48kHz. It is easier to notice the small things in detail, and the sound is "cleaner defined"if that makes sense, provided your playback equipment is good enough.
I do have difficulty noting audible differences between 48kHz and 96kHz sample rate though. Just use 96kHz anyway because it has higher potential, filesize is not a subject these days anyway.
Other than that I agree, bit depth and sample rate frequency is the new watt, "more = better" apparently.
Last edited:
most Studios already work at 24/96 - so resampling with dither are standard plugins in their mastering software - pretty universal, has to be used to produce 16/144 releases - little "new" education required
to my knowledge the few published peer reviewed articles claiming positive results in controlled listening for higher resolution in either bit depth or samplerate than 16/44 haven't been replicated, have methodological questions, aren't yet accepted as proven psychoacoustics
but I'd say the situation is unsettled and therefore boosting sample rate is a better use of any extra bits - little of the current sophistication in the dither algorithms is required at much higher sample rate - the noise doesn't have to be so carefully apportioned in our normal hearing range - it can all be above 20 kHz with 96 k
If you can reliably tell 24 vs 16 bit depth, when a good dither has been properly applied to the 16 bit data, with music at reasonable volume then there are psychoacoustic researchers wanting to test you in their labs
to my knowledge the few published peer reviewed articles claiming positive results in controlled listening for higher resolution in either bit depth or samplerate than 16/44 haven't been replicated, have methodological questions, aren't yet accepted as proven psychoacoustics
but I'd say the situation is unsettled and therefore boosting sample rate is a better use of any extra bits - little of the current sophistication in the dither algorithms is required at much higher sample rate - the noise doesn't have to be so carefully apportioned in our normal hearing range - it can all be above 20 kHz with 96 k
If you can reliably tell 24 vs 16 bit depth, when a good dither has been properly applied to the 16 bit data, with music at reasonable volume then there are psychoacoustic researchers wanting to test you in their labs
Last edited:
Far be it from me to suggest that anyone would want to visit anywhere other than diyaudio, particularly somewhere as regularly derided as hydrogenaudio, but anyone who did would discover plenty of solid theoretical and practical foundation to suggest that despite numerous attempts to demonstrate otherwise, no audible improvement over well-engineered 16/44k1 is achievable.
Considering the difficulties encountered by those determined to prove the necessity of higher bit depths/sample rates, it has to be acknowledged by any fair-minded pundit that if any improvement is possible, it must be absolutely marginal.
It might be comforting to think that the system one enjoys exceeds the limits of audibility by some margin, but it's a disappointment to me to see a creeping acquiescence to a mass-market sales strategy here on what I take to be an evidence-based forum aimed at home constructors.
Greater bit-depth in a DAC may have some justification in terms of allowing digital (firmware) attenuation, but the recordings we can buy on CD already are high-definition, if there are problems with dynamic range they are the result of choices made by recording engineers not system engineers and there is no reason to think that a technically more capable system will improve matters.
Hopefully I don't put my foot in my mouth, but why would you attenuate in the digital realm? You would lose dynamic range no matter how many bits there are. Bits determine the resolution of the signal and should not be used to "control" volume or dynamic range. Example - 4 bit system - 0000 = lowest level, 1111 = highest level, only 16 distinct steps between them. 8 bit system - 00000000 = lowest level, 11111111 = highest level, now there are 256 distinct levels between them. There are no negative levels, so if you attenuate digitally, the number of levels between low and high decrease. Seems to make sense to leave the digital alone, and attenuate in the analog realm - which isn't that difficult.
In other words, if you attenuate 50% digitally, everything below 50% is gone. The way to keep it all is to compress downwardly, causing the high level to be 50% less loud, but again, the upper 50% of your dynamic range is not being used.
Make sense? I deal a lot with vibration signals in the digital world and while I may not have done the best job conveying the message, I do know what I am talking about.
In other words, if you attenuate 50% digitally, everything below 50% is gone. The way to keep it all is to compress downwardly, causing the high level to be 50% less loud, but again, the upper 50% of your dynamic range is not being used.
Make sense? I deal a lot with vibration signals in the digital world and while I may not have done the best job conveying the message, I do know what I am talking about.
Well engineered is the sore point here methinks. And I am not certain at all if it's the bit depth that makes the difference or how the bits are used. Maybe if I had ACTUAL 16 bit and 24 bit it would be impossible to distinguish. Have no real knowledge about this, but I seem to remember reading a thread in the digital section here, where they argued that hardly any 24bit dac's are actually 24bit, more often between 18-21bit real bit depth. So following that line of thought, maybe a 16 bit dac is not 16 bit, but more often 13-15 bit?
most (even audiophile) home playback will be through systems that clip between 100 and 110 dB SPL, in rooms seldom reaching NC25
so in rough numbers 95 dB S/N would be just about the most available
so you just need 16 bits when playing at very loud volume – any turn down in the above situation will have the room noise floor as a limit – volume control technology doesn't matter
and the “lost bits” heuristic itself is past its “use by” date – today Media player SW calculates with 32 bit internal wordlength, can be set to send 24 bits to your HD DAC – even with 16 bit source
so in practice today even with 16 bit source the bits don't "get lost" until they are way below better grade audio DAC output electronic noise floor ~ -120 dB
and again it is hard to get through to people what dither does - you can have much more than 16 bit's theoretical 96 dB when the dither is shaped to hide where our hearing isn't sensitive - delivering weighted psychoacoustic S/N well over 100 dB
so in rough numbers 95 dB S/N would be just about the most available
so you just need 16 bits when playing at very loud volume – any turn down in the above situation will have the room noise floor as a limit – volume control technology doesn't matter
and the “lost bits” heuristic itself is past its “use by” date – today Media player SW calculates with 32 bit internal wordlength, can be set to send 24 bits to your HD DAC – even with 16 bit source
so in practice today even with 16 bit source the bits don't "get lost" until they are way below better grade audio DAC output electronic noise floor ~ -120 dB
and again it is hard to get through to people what dither does - you can have much more than 16 bit's theoretical 96 dB when the dither is shaped to hide where our hearing isn't sensitive - delivering weighted psychoacoustic S/N well over 100 dB
Last edited:
There is validity in that argument. In my job, if I use a vibration sensor that outputs a level from 0 to 500G, but my test only covers 10 to 250G, then I am not using the full resolution, or bit depth of possible levels. A sensor that outputs 0 to 250G would be much more appropriate. But how do you translate this to music? If your highest level is 100dB, but most all the rest of the content runs at 65dB, you rarely realize the full bit depth available. Just the way it is...
Yes, free samples of "HiRez" are easy to find. However, I don't find much, if any, that actually specifies that no compression was used.
People want to be able to move a slider on their computer screen and analog attenuation isn't built into most digital systems.
Some modern DACs have built-in digital attenuation schemes operating under uProc control. It's of course possible to construct a digital-controlled attenuator operating solely in the analog domain, but it's not always done.
It's difficult to know the exact implementation details of every piece of equipment one encounters. I always run the digital end at maximum volume, it's the simplest way to the best result. Hopefully. It's a kind of diminishing returns thing, there's a limit to the amount of personal effort I want to put into playing some music, particularly if I don't notice a problem, and all my systems have regular pots in them.
This is where I disagree. the difference is impossible to notice when running full tilt constantly, it is in the transitions and quiet passages with low level details the difference appears. Like if you are playing a track loud, all of a sudden there is a transition to a quiet passage, you are listening to the noise floor of the guitar amplifier on the recording, and the guitar player is playing very very carefully, you can hear the uncut/not close cropped strings moving ever so slightly, the musicians breathing very quietly, someone in the room just opening their mouth without saying anything maybe just a smile or grin. This sort of thing. Just love those recordings, they are rare, but great moments in recording history.
Now, I agree that proper 16bit recordings are really good, but they are far between and the so called experts in the studio are often too lazy to do things properly, they may not use optimal equipment or maybe they use it wrongly, but resolution and bit depth is never directly wasteful. There may come a day when everyone is able to hear the difference and truly appreciate it.
Right.
But it's a 'weakest link in the chain' scenario. No amount of beefing up the other links helps.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.