You seem to be convinced that one can't hear differences, and it seems that you are one of those that are confused by the difference between bit-depth and sample rates. You seem to be talking about sample rates. Again: do your own testing if you have the means I described: dither on your computer with software that will play true 24-bit sound, and do your own tests dithering a 24-bit file down to 16-bit using various dither schemes. That will train you to hear the differences. My software allows me to do direct A/B comparisons between the source and various dither schemes. It also appears that some people are conflating bit-depth with mastering because they do not understand what is being discussed.
You're right, typically if a person believes something and feels it's their truth, then it's hard to change minds. Suppose it's true that you can hear a difference between 16 and 24-bit files because you did this with the volume controlled and in a reasonably blinded fashion to reduce psychological bias, then yeah, system quality could play a role. Realize that "system quality" needs not imply that it's because the system is "good". Maybe the DAC performs some kind of processing on the 16/24 bits differently. Maybe there could be slight output level differences affecting the audibility. Maybe the dithering algorithm to 16-bits was poor when running the A/B, or the change just truncated bits. Maybe different mastering was used. Maybe different filters are applied (eg. rarely do we see 16/96 content but plenty of 24/96, so a device like the old Pono Player will apply different filters between 48 and 96kHz). While humans have idiosyncrasies, we're not aliens, and are closer to what empirical research tells us than sometimes what we're willing to admit - especially since the vast majority of us here are males, >30 and likely into midlife. Outlier individuals in hearing ability to the extent of telling the difference between say a 24/48 vs. the same file dithered to 16/48 using a good TDPF technique at normal listening levels in normal rooms starting with true hi-res audio I think would be very few. It would be much easier with headphones using dynamic tracks with very low noise floor like a very good classical recording. What I often see among friends are some who claim they have extraordinary abilities to hear all kinds of things automatically assume that it's because other people have poorer sound systems and poorer hearing. Assuming they did their own listening test correctly, it's just as likely that the expensive "high end" stuff, and expensive supposedly "hi-res" downloads add their own distortions and anomalies, and maybe that's the cause of what's being heard. (As I discussed here, there are hi-res downloads will all kinds of ultrasonic anomalies that could actually make them sound worse or at least different depending on the gear.) In my life and among audiophile friends here, I have seen a number "convert" to more realistic viewpoints about this stuff as they understood more and earnestly performed their own testing over time. So while few minds might be changed in a thread like this, I think it's important to try and discuss these things honestly while employing the scientific data we know about humans and these technologies. Over time, I believe there is an impact... No mud needs to be slung.
No, I am not confused at all. Bit depth determines dynamic range and noise floor. Sampling rate determines upper frequency range. We are talking about 24 bit versus 16 bit. So noise floor.
That happened to me. Many things I assumed or even thought I could hear, were proven NOT SO AUDIBLE when not sure of the source............ I would rather be humble and unsure, than outspoken and say I can hear everything. A bit more humble helps.
Maybe let us know what music you're using. What software you use. And what listening level you're testing... Sure, I can hear the 16-bit dithered noise floor +/- noise shaping using very low level sounds and very high gain settings using my THX AAA headphone amp and Sennheiser HD800. But anyone with decent hearing and a reasonable sound system can do that and I wouldn't count that as a realistic nor meaningful for normal music and listening levels!
Good reply. Sure anyone can "Force" a sound to get to the levels to tell, but with actual musical content, another story.
By definition 0 dBFS is the loudest possible digital signal. 16 bit can resolve details down to - 96 dBFS 24 bit can resolve details down to - 144 dBFS Indeed 24 can reproduce micro details below -96 so is superior to CD but...... - you need a very, very quiet recording chain. If so, expect musical life down to -120 at its best - most of the time 24 bit recording contains just a litle bit more e.g. down -100 dBFS making it hard to distinguish them from a CD - you need a very quiet playback chain. If your power amp has a SNR of 80 dB, you might wonder if you can even hear the details of a CD. - modern power amps often have 100 dB as SNR. So a lot of details of a good 24 are buried in the noise floor. - amps based on Eigentakt or Ncore modules might come close to 120 dB SNR and are able to fully resolve a good 24 recording but I'm afraid you have to play FFF loud to retrieve thos tiny details.
I've not tested it but will take John Siau's word about it. He's they guy that designs Benchmark's DACs. A while back I read a discussion on HD-Audio where he said all we need is about 18 bits for playback. Based on his experience, between the two choices, 16 and 24, I'll shoot for 24.
Yup, good reminder Vincent. By SACD Scarlet Book standards, peak level should be "0dB DSD", equivalent to -6dB FS in the PCM world. And absolute peak "cannot" officially go above +3.1dB DSD or -2.9dB FS. Although some SACDs have neglected this over the years...
FWIW, it gets all the other combos right, including 24/48. My guess is that it’s actually playing 24, but the display isn’t registering it.
I use MBIT+, and the software I use utilizes it. I use primarily iZotope, which is industry-standard software. (plugins, DAW). I don't listen to or worry about the noise floor. I listen to the effects the various dither schemes have on the music itself. I listen to all types of music.
The BIG question, when you say you can hear the difference, is: "have you done these tests BLIND, that is with someone else playing the file for you?". Because if you haven't, and you KNEW which file was playing, there is a simple word for it: placebo effect.
I would not go "that far", but I would say placebo or expectation is possible, or at least could factor in.
Same thing happens on Amazon. Doesn’t bother me. Could be the Marantz , but it could be the record companies.
The software I use allows you to do A/B blind testing with a bypass button, even though it's not necessary. It doesn't matter if you know what the source or dithered version is. The goal is to hear what the dither is doing to the sound, not if you can hear differences. I assume you have never compared dithered files against their sources because you are hell-bent on believing that there aren't any differences. I've been doing this stuff for 34 years. I said "this again" in my first post because these threads always...ALWAYS degenerate into people like you deciding for everybody about what they can and can't hear and can't be bothered to test anything themselves.
I’ve had fun experimenting with that too, via my Meridian 518. If anyone wants easy push-button access to various dither/noise-shaping methods, as described in page 16 and 36 here https://www.meridian-audio.info/public/518user[120].pdf I can recommend it. It also makes it easy to perform single-blind tests with a friend.
Actually I have! And I have subjected friends to the same blind test! Complete 50/50 randomness in the choices! And also a very funny test to debunk a friend who swore on his mother that he could hear a difference between a CD and its bit-perfect copy ("there is less bass in the copy!!!"), but when he was subjected to the blind test he got all the answers wrong. :-D
Since random would be 50/50, getting them all wrong means he could clearly hear the difference. He just had the “sign” wrong.
One thing I always wonder is: yes, 24 bit gives in teased dynamic range, but does it not also give smaller steps in volume? If not, why not?