[1] The guantlet has been baseless accusations aimed at debasing me & discrediting me as a person -
[1a] that is not part of the "scientific method."
[2] I have not presented any false information.
[3] I used standard tools (sox) in the manner recommended & disclosed my config parameters in case they needed any adjustment. I made no modifications beyond that. If you want to provide the 16bit version you made I can retest.
[4] I did not change the volume nor did I use any tools to measure / display the bit rate or sample rate. I just used my ears and the difference is very obvious to them.
[5] What is silly is that some who "claim" to adhere to science are really more akin to religious zealots when presented with evidence that may call into question their beliefs.
1. I personally have NOT debased or discredited you directly as a person, although I have made an accusation aimed at discrediting/debasing the information you've presented. However, it is
NOT a baseless accusation, it is based on an objective measurement, a null test which demonstrated a null of only (very roughly) 20dB.
1a.
Absolutely it is, in fact it's arguably the most important part of the scientific method!! The scientific method absolutely requires the very close scrutiny of ALL aspects of a test/experiment by others: The test material, the test methodology, the analysis of the resultant data and the asserted conclusions from those results. In fact, the scientific method REQUIRES at least two steps in this process: Firstly, peer review and then replication of the results by others repeating the test/experiment. Faults/Errors are frequently found during this process and depending on their seriousness, the end result can easily be that the scientist/s are personally discredited!
2. Clearly you have presented false information. To start with, as already mentioned, there is some error in the down converted file you've used for comparison. Although again, I'm NOT accusing you of deliberately presenting false information.
3. But you haven't "
used standard tools in the manner recommended"! You have presented a commercial music mix and the commercial standard for creating a 16bit version from a higher bit depth original/master is NOT sox, nor the use of standard (TPDF) dither. The "recommended standard" is the use of noise-shaped dither and there are/have been a number of tools used as "standard" over the last 25 years or so, for example, in order of time period: Sony Super Bit Mapping, Apogee UV22HR, Pow-R, iZotope and various others but not sox. However, the developer community does appear to use sox quite commonly, presumably because they are thoroughly comfortable using a command line interface. My guess (and it is just a guess because like most audio engineers I've never used sox), is that you've made some error in the parameters you've defined, because I've be very surprised if the developer community had missed some bug or deficiency in sox.
4. There are various reasons how/why a difference 16 vs 24 bit can be detected. I've already mentioned an "unreasonable" listening level but there are other possibilities, for example: You are also downsampling and it's possible that the original contains enough >22kHz content to cause audible IMD in your reproduction chain, while of course the 16/44 version would not contain any >22kHz content. There are other possibilities, such as incorrect settings selected in Foobar's the ABX plugin. The problem is that you are claiming "very obvious" audible differences and "very obvious" audible differences require relatively large objective differences. However, digital audio theory does NOT predict such large differences, this is confirmed by actual objective measurements and further confirmed by countless thousands of controlled listening tests over a period of 25 years which demonstrates no audible difference at all and certainly NOT a "very obvious" audible difference. This extensive body of interlocking supported evidence represents well established, accepted science, which your claim contradict! And, whenever such a claim has been made here, it has ALWAYS turned out to be an error in testing methodology and/or materials, sometimes an inadvertent error and sometimes a deliberate fraud.
5. There two rather obvious problems/falsehoods with this assertion: Firstly, you are NOT "calling into question our beliefs", you are contradicting a very well established, accepted body of evidence/science. And secondly, it's a basic axiom of science that "
An extraordinary claim requires extraordinary evidence". However, you have not presented extraordinary evidence. You have attempted to provide evidence that is more reliable than the sighted tests typically presented by audiophiles, however, there are serious questions regarding your two comparison files and your methodology, so you have not yet reached the bar required even for reliable evidence, let alone extraordinary evidence! Until you do, extreme scepticism is the ONLY logical response from those who "adhere to science", which is pretty much the opposite of "silly"!!
[1] So this is the reason I can hear the noise floor of many recordings, I know it's this because all songs in an album have a particular hiss, and if I start listening to another one, the hiss is different.
[1a] It has being annoying to say the least.
[2] I listen at around 80 to 90 dB SPL with headphones.
1. But that's observational evidence that you are NOT hearing the digital noise floor (dither)! This is because dither does NOT change, it's of constant level (equivalent to about the LSB) and constant frequency/spectral distribution (pure white noise). In other words, if you were hearing the digital noise floor (dither), then the hiss would NOT be different. However, both the acoustic noise floor (the noise floor of the recording venue) and the analogue noise floor on a digital recording are very highly variable both in level and spectral distribution. In fact, the acoustic noise floor varies very audibly even in the SAME recording venue, due to slight differences in relative mic positions and slightly different locations within the recording venue, which presents particular problems/difficulties with TV/film sound. Bare in mind that the noise floor of standard dither is about -92dB, while the recording noise floor (acoustic + analogue noise) of commercial music recordings is typically between -40dB and -60dB.
1a. Different songs on an album will have somewhat different playing styles, often different instruments and/or different instrument settings, will have been recorded and mixed somewhat differently and will therefore have a different acoustic (and probably a different analogue) noise floor. There's not much that can be done about this.
2. That's very loud, unless:
A. You have headphones that do not isolate much and you use them in a relatively high noise floor environment or
B. If you're listening to material that rarely reaches peak level, only for very short periods and the vast majority of the recording is significantly lower than peak, EG. Certain relatively uncommon pieces of classical music, that have not had any audio compression applied.
Even though it's probably uncomfortably loud for most people, 90dB SPL is still within "reasonable listening levels". With 16bit, using standard dither, the digital noise floor would be about 92dB below peak, so if your peak is 90dBSPL the dither noise would be at about -2dBSPL. Bare in mind that a 100w incandescent light bulb produces about 4 times more noise (~10dBSPL at 1m)! Even with a somewhat higher listening level, world class anechoic chamber conditions, a hypothetical recording with an acoustic + analogue noise floor lower than -92dB and an amp/speakers or HPs that have a combined noise floor lower than -92dB, you'd still be very near the threshold of audibility. However, a somewhat higher listening level (than 90dB) would be very close to an ureasonable listening level in world class anechoic conditions AND, recordings with a low noise floor would/should ALWAYS have noise-shaped dither applied (putting the dither noise at about -120dB), so the dither noise is well below audibility even at the hypothetical/theoretical extremes of "reasonable listening levels" (which of course is why it was invented in the first place)!
Still think you're hearing dither noise on commercial 16bit music recordings reduced from 24bit?
G