71 dB
Headphoneus Supremus
Regardless of meaning how does the same track sound 24/16 ? For me it's obvious that redbook missed the Boat at using 16 . Any same track at 24 shows a much lower noise floor
It plays from blacker background.
Even 24/88.2 is miles better.
The lower noise floor maybe due to a much higher band width some 256 more in dynamics.
So asdide from math does anyone hear this improvement ?
13 bits
What is needed in consumer audio is 13 bits worth of dynamic range (about 80 dB). CD is kind of overkill already, but only 3 bits so who cares? 24 bit audio is technically 11 bits overkill, but in practice a lot less than that, because of the lowest 8 bits most if not all is noise unless the sound is completely computer generated (so, Autechre's 24 bit files might actually have 24 bit worth of dynamic range, but of course you can't really experience it for many reasons and the same files dithered to 13 bits would sound the same).
If you hear the noise floor of 16 bit audio, the reason is:
(1) The recording contains high levels of background noise. 24 bit version would have the same noise.
(2) Your volume is turned to insanely high levels to hear the noise floor. You would never listen to music at those insane levels.
Dither expands dynamic range below the LSB, in fact there is no limit to that other than the fact that the dither noise masks the signal depending on the type of dither. Noise shaped dither noise mask less and can expand the perceivable dynamic range even 20 dB! Quiet sounds do not granulate (modulate the noise), but they "live" in the noise like signals in analog audio.
How "loud" is dither noise? You can generate TPDF dither in Audacity following these steps:
STEP 1 - Open Audacity and select Generate ---> Silence ---> 10 seconds
STEP 2 - Select Audio Track ---> Set Sample Format ---> 24 bit PCM
STEP 3 - Select Audio Track ---> Set Rate ---> 44100 Hz
STEP 4 - Select Effect ---> Nyquist Prompt ---> write (mult (sim (noise) (noise)) (recip 65536)) and click ok
Try to hear the noise. The level meters show there's stuff at level around -90 dB. Turn up volume until you hear the noise. Now, listen to a music file using the same volume setting if you dare! Don't do it, because you probably damage your ears and gear! That's how quiet TPDF dither is. Noise shaped dithers are perceptually even quieter!
44.1/48 kHz
No higher samplerate is needed than 44.1 or 48 kHz in consumer audio. Those cover the human hearing range 20-20000 Hz in the childhood (upper limit lowers with age). However, if the music is for dogs (a dog whistle concerto), 88.2 kHz is kind of the lower limit I would use, 96 kHz would be safer as dogs can hear up to 45 kHz I believe. Cats can hear up to almost 80 kHz, so their audio formats would probably use 176.4 kHz or 192 kHz sampling rates. Dolphins would need to use 352.8 kHz!