Is big depth related to anything but dynamic range? If not, I’d be curious to hear a mathematical explanation of why 24 but would be better, since the dynamic range of almost all recordings can be conveyed easily with 16 but.
IME with 24 bit recordings (purchased on CD then ripped to server) the presentation seems to be a little hotter with more air around instruments (reverb/decay transients) when compared to typical Redbook audio but only when switching from one to another without level adjustments. Once level matched I think it would sound like any other well mastered recording.
Yes, bit depth determines the dynamic range. Noise floor to the loudest sound it can reproduce. 16 bit already covers 96db of dynamic range. The best analog tape was less than 80db of dynamic range so 16 bit as the final presentation should be enough. For mixing and mastering, going 24 bit gives you more headroom for calculations if you need to make digital eq and volume adjustments. But if you all the mastering is done analog (as Steve does), you can record to 16 bit as the final product because you aren't make any eq or volume adjustments.
That's the real issue. Most people aren't truly comparing 16 vs 24 bit with everything else equal. Having a CD of something and then downloading a hi-res version doesn't make a 16 vs 24 comparison. What does is taking a hi-res file and down converting to both 16 and 24. Then compare that.
This again? Yes! There is a difference, and many of us can hear it. If you can have a 24-bit file and have the means to dither it to 16-bit yourself using good software, compare the various settings, and compare them to the original, you can hear the differences, and learn what to listen for. The better the dither, and its settings, the harder it gets to distinguish between the source and the dithered result. I use MBIT+. I imagine most people who claim not to hear any differences do not have the means to dither.
Agreed, but if you aren't starting with more than 24 bits of resolution and using the same filtering algorithm to down-sample to 16 bits and 24 bits, then you have the differences in filtering algorithms to contend with. As others have said, it's MUCH more about the quality of the recording and mastering than it is about the resolution. Of my hundred favorite songs from a sound quality perspective, most are standard 44.1/16bit.
Do you go by what the majority says, or by what you can hear, because, despite what some people say, it's your ears that matter, not mine or anyone else's. But, if you have the means, do your own tests. Your personal results should be your answer for yourself.
The short story is with everything functioning properly, and not applying DSP, you will hear NO difference between 24 and 16bit for end user playback. …and sometimes there IS….because of the same reasons.
If we're just comparing bit-depth, then I guess the file should remain in its sample rate. So I'd say take a 96/24 file and dither to 16 and compare that to the original 24 bit file. If a quality program was used, I am still confident that they would be inaudibly different on blind test. Would be interesting to compare the different types of dither and even truncation to see if blind tests reveal any of those to be audible. I wouldn't be surprised if people can't tell the difference between truncated 16 bit vs 24 bit. Still, even if we are downsampling from 96 to 44.1, we'd be doing it for both, so there shouldn't be a difference in filters. And actually, the resample process would put both files in floating bit so both would then have to be dithered or truncated to get to 16 or 24 bits. So this might actually be the better test, But I think either would do.
At 32-bit (24-bit with an 8-bit mantissa), you can exceed full-scale decobel 0 to your heart's content. Just remember to bring it back under 0 before you archive or take it down to 16-bit. 32-bit will not clip. As far as leaving a cushion is concerned, if you ever want to make mp3 copies, it's a good idea to leave a full 1 db. I personally leave -.6 db because I know exactly how much my mp3 LAME encoding at 320 kbps CBR will boost the peaks.
The primary advantage of higher bit-depth (for the final format) has relatively little to do with the lower noise floor of 24 bit PCM vs 16 bit PCM. It has to do with the fact that 24 bit PCM by definition has 256 voltage-levels per each 16 bit voltage stair-step. And each time you double the sampling you get double the time resolution. The combination of these results in a dramatically smother waveforms when you approach the least significant bit of 16 bit PCM signals. For those who claim that these stair-steps are purely a mythological creation, here is a capture from an oscilloscope showing exactly what an undithered sine wave looks-like when 16 bit PCM approaches its least significant bit.
I imagine the answer to this question is dependent on how resolving one's system and ears are. The question isn't whether 16 bit files or CD's can sound good, they can. I look forward to testing out the difference soon with my new SACD player next week.
Good point on the filtering algorithm. I can hear an improvement with 24 bit, but like you said, most of my highres sounds terrible and most of my best sounding albums are redbook. But, for the good sounding highres I have, it sounds better than 16/44.1 when I have experimented converting it to 24/44.1 and 16/44.1. Not all 24 bits are needed to answer the thread’s question, if even 17 bits are better than 16 bits.
Oh, I agree and then some. I think though that OP is referring to the bit depth of the source recording. As someone noted the greater bit depth results in more dynamic range. I decided to look at an opinion about whether the difference would be audible. "Not exactly," says The Hive. 16 Bit vs 24 Bit (Clearing Up The Confusion)
24bit recordings sound better played back at 24bit than 16bit. The sound difference to look out for is the 16bit version doesn't 'swing' as well as typified in this track. (Easy to compare 16bit v 24bit on Qobuz):
Rather, that is digitally-generated 16 bit audio, played on a 20 bit R-2R DAC, recorded by a 24 bit ADC... How about instead, the appearance of music approaching the last bits in quiet passages, quantized to discrete and visible values: That is 8 bit audio, centered on 128, encoded between values of 0 and 256. Surely one must be able to distinguish the "grit and graininess" with audio downsampled to 8 bit? Have a listen to what you see: brothers-16bit.flac brothers-8bit.flac The top original has by definition 255 voltage levels between every eight bit step...
Here’s the difference file from a 24/88 vs. 16/88 null. Dropbox - PSUCCDelta2488v1688_DAAD.wav - Simplify your life
Fun fact: 24 bit offers dynamic range of 144db, which is like going from the silence of space to the sound on the deck of an aircraft carrier. If you start from the 40db of a “quiet” room, it’s like going from that room to the sound of a gun fired within a few inches of your ear. 32 bits allows us to go from a quiet living room to the sound of Krakatoa from a few miles away. Basically what I’m saying is, start saving for those monoblocks.
Nah, even with $50k of gear, I doubt it. And don't try it with young guys in an anechoic chamber just in case. ;-) Couldn't find a significant blind listening test difference with audiophiles using genuine 24-bit 2L recordings back in 2014: 24-Bit vs. 16-Bit Audio Test - Part II: RESULTS & CONCLUSIONS The science has always been quite clear that hearing a difference between 24-bits and 16-bits when music is encoded to utilize the dynamic range afforded by 16-bits is minimal. Nonetheless, I still like my favourite music in 24-bits if I believe it's sourced from hi-res. And when dithering, also will use MBIT+ dithering since a little bit of noise shaping to improve dynamic range in the most sensitive frequencies (say 1-6kHz).
The expected result should be the higher 16-bit noise floor if that is the difference between the two files. You can't "null" to probe signal below the highest noise floor of the two files since that noise will be the result and will swamp any signal beneath it (were it to exist).
