When I see the audiophile versions, I wonder if anyone does really have the patience to download a 24/192 .flac, if they could have pretty much the same perceived quality with standard 16/44.1, considering they don't have any large speaker system like those in theaters or at concerts. The CD quality of 16-bit and 44.1 kHz was devised exactly to sound clear enough to human ears and the diskspace is now so cheap it can be stored even uncompressed at all.
Samplerate
Typical human ears can hear frequencies from 20 Hz to about 20 kHz with high-reject filtering level increasing along with age, capping the audible range at something like 15 kHz at the life expectancy age. That means the reasonable sampling rate is above 40 and below 80 kHz. The 44.1 and 48 are high enough even for kids and elderly people could accept 32 kHz. The only downside of this loss is children have been very annoyed if their (grand)parents forgot to turn off the CRT TV completely, leaving it tweeting at 15 to 20 kHz depending on the particular model, until the LED TVs have become cheap. (Plain LCD sucks for a TV, because the picture looks worse than at CRT and TV is not made to be used as a monitor for a computer.) If you don't know why your <2 y/o child screams, check if you haven't forgot to turn of the CRT TV if you haven't already upgraded to a totally useless 4K, on which pixels blur as you lean like 6" away. This is again a case of excessive sample rate, this time in a picture. Apple's retina display is a waste of resources too, because the higher resolution won't compensate for relatively fat fingers of not only obese people, who keep missing the buttons of the software keyboard. The same goes for sound, where 48 kHz is enough when using headphones and 16 kHz is an excess if using the integrated cellphone speakers, which forget to play the bass unless held annoyingly close to an ear. For samples you might want to record in 96 kHz if you're going to downtune them a octave, that means to play them with a samplerate of 48 kHz. Don't set excessive samplerates for the mixed output if you know the final product will be 44.1 kHz anyway. Ripping a vinyl is different than synthesizing music, therefore 96 kHz is acceptable, while 176.4 and 192 kHz is still a waste and it only introduces some ultra-sound noises. Consider it like archiving for aliens, which might have a way better hearing than us so psychoaccoustic algorythms of lossy encoders won't work on them. The digital sampling is dull by design, 96 kHz won't make it sound better in any way.
Bit depth
The dynamic range of 16-bit sample resolution is 96 db. For 24 it's 144 db. For our technology, it's at most about 20~22 usable bits (if the metal parts were all gold, it wouldn't help much) which is why there exists a 20-bit compromise in the Super Hi-Vision specification. And there are our ears, which start to hurt as the over-compressed music for "everyone's" speakers starts to exceed 130 db. And all the engineers working near a heavy machinery know they need to put something in their ears if the noise is exceeding 85 db for a standard 9-to-5 shift. That means, even the CD has enough dynamic range to affect your hearing in about 6 hours when adjusted to one level meaning a barely audible change. Is there still any point in heaving such a large headroom with 24 bits when you could get earraped by a random error sound? Of course there is a scenario, where 24 and 32 bits are useful, and that is the processing. There has to be a margin for all the rounding errors that occur when amplifying the signal and 32-bit floating point precission provides it. Once the final master is made, dithered 16-bit is more than enough, providing up to maybe 120 db of dynamic range if made properly.
Conclusion
Because of human hearing, a higher bit depth is relevant for some spaces (definitely not home), but the sample rate is definitely not. Considering the speed of your Internet connection and personal use of all the pirated music, stick with 16/44.1 .flac and you're fine. If you are making some music, use 32F/44.1 (16/96 when ripping vinyls, 32F/192 or 64F/384 when studying the artifacts your hardware makes) as source format for all the masters for specific carriers and maybe get quantized 24/44.1 files on a SACD, but if selling the music through internet, 16/44.1 will do better than 24/96. Who has the time to download trice as much data, when they are not going to hear like 2/3 of them? Clearly no one at least a little normal unless they have some unused 8TB harddrives or a pet dolphin, whale (300 kHz samplerate), bat (250 kHz), mouse, cat, rat (200 kHz), hegdehog, sheep, cow, horse and many others, but definitely not tuna, chicken, fish, frog, elephant and owl, which they want to enjoy the music too.
This is also related: https://people.xiph.org/~xiphmont/demo/neil-young.html
Samplerate list (SEO)
1000 Hz
2000 Hz
3000 Hz
4000 Hz
5000 Hz
6000 Hz
8000 Hz
8363 Hz - Amiga low C
10000 Hz
11025 Hz - Soundblaster
12000 Hz
16000 Hz
16726 Hz - Amiga high C
20000 Hz
22050 Hz - SoundBlaster Pro
24000 Hz
32000 Hz
33075 Hz
40000 Hz
44100 Hz - CD
48000 Hz - DAT
64000 Hz
66150 Hz
88200 Hz
96000 Hz - Hires audio
128000 Hz
176400 Hz
192000 Hz - Artifact study audio
256000 Hz
352800 Hz
384000 Hz - Dolphin audio
512000 Hz
705600 Hz - lowest common multiple of 44100 and 48000, used for resampling
etc. - 32000*2*2*2*2*2..., 48000*2*2*2*2*2..., 44100*2*2*2*2*2...
2116800000 Hz - 44100*48000, divided by 2 until it can't gives 4134375 Hz.
14112000000 Hz - lowest common multiple of 44100, 48000, and 32000. Divided by 2 until it can't gives 6890625 Hz.
From Facebook:
kHz: 8, 11, 12, 16, 22, 24, 32, (33, 40,) *44*, 48, (64, 66,) 88, 96, (128,) 176, 192, (256, 353, 384)
bps: 8, 10, 12, 14, *16*, 20, 24, 32, (64, a-law, u-law, 16F, 32F, 64F)
chn: 1: 1.0, 1.1; *2: 2.0, 2.1, 3.0, 3.1*; 4: 4.0, 4.1; (6: 5.1, 6.0, 6.1; 7.1, 22.2)
mode: PCM, (delta, deltabit,) DSD, unsigned, FLAC, (WV, WMAL, ALAC, ZIP, RAR, 7Z)
6.0 a 6.1 - each speaker on 1 side of a cube. 3.1 can be transmitted as stereo with 1 center channel playing both.
Samplerate
Typical human ears can hear frequencies from 20 Hz to about 20 kHz with high-reject filtering level increasing along with age, capping the audible range at something like 15 kHz at the life expectancy age. That means the reasonable sampling rate is above 40 and below 80 kHz. The 44.1 and 48 are high enough even for kids and elderly people could accept 32 kHz. The only downside of this loss is children have been very annoyed if their (grand)parents forgot to turn off the CRT TV completely, leaving it tweeting at 15 to 20 kHz depending on the particular model, until the LED TVs have become cheap. (Plain LCD sucks for a TV, because the picture looks worse than at CRT and TV is not made to be used as a monitor for a computer.) If you don't know why your <2 y/o child screams, check if you haven't forgot to turn of the CRT TV if you haven't already upgraded to a totally useless 4K, on which pixels blur as you lean like 6" away. This is again a case of excessive sample rate, this time in a picture. Apple's retina display is a waste of resources too, because the higher resolution won't compensate for relatively fat fingers of not only obese people, who keep missing the buttons of the software keyboard. The same goes for sound, where 48 kHz is enough when using headphones and 16 kHz is an excess if using the integrated cellphone speakers, which forget to play the bass unless held annoyingly close to an ear. For samples you might want to record in 96 kHz if you're going to downtune them a octave, that means to play them with a samplerate of 48 kHz. Don't set excessive samplerates for the mixed output if you know the final product will be 44.1 kHz anyway. Ripping a vinyl is different than synthesizing music, therefore 96 kHz is acceptable, while 176.4 and 192 kHz is still a waste and it only introduces some ultra-sound noises. Consider it like archiving for aliens, which might have a way better hearing than us so psychoaccoustic algorythms of lossy encoders won't work on them. The digital sampling is dull by design, 96 kHz won't make it sound better in any way.
Bit depth
The dynamic range of 16-bit sample resolution is 96 db. For 24 it's 144 db. For our technology, it's at most about 20~22 usable bits (if the metal parts were all gold, it wouldn't help much) which is why there exists a 20-bit compromise in the Super Hi-Vision specification. And there are our ears, which start to hurt as the over-compressed music for "everyone's" speakers starts to exceed 130 db. And all the engineers working near a heavy machinery know they need to put something in their ears if the noise is exceeding 85 db for a standard 9-to-5 shift. That means, even the CD has enough dynamic range to affect your hearing in about 6 hours when adjusted to one level meaning a barely audible change. Is there still any point in heaving such a large headroom with 24 bits when you could get earraped by a random error sound? Of course there is a scenario, where 24 and 32 bits are useful, and that is the processing. There has to be a margin for all the rounding errors that occur when amplifying the signal and 32-bit floating point precission provides it. Once the final master is made, dithered 16-bit is more than enough, providing up to maybe 120 db of dynamic range if made properly.
Conclusion
Because of human hearing, a higher bit depth is relevant for some spaces (definitely not home), but the sample rate is definitely not. Considering the speed of your Internet connection and personal use of all the pirated music, stick with 16/44.1 .flac and you're fine. If you are making some music, use 32F/44.1 (16/96 when ripping vinyls, 32F/192 or 64F/384 when studying the artifacts your hardware makes) as source format for all the masters for specific carriers and maybe get quantized 24/44.1 files on a SACD, but if selling the music through internet, 16/44.1 will do better than 24/96. Who has the time to download trice as much data, when they are not going to hear like 2/3 of them? Clearly no one at least a little normal unless they have some unused 8TB harddrives or a pet dolphin, whale (300 kHz samplerate), bat (250 kHz), mouse, cat, rat (200 kHz), hegdehog, sheep, cow, horse and many others, but definitely not tuna, chicken, fish, frog, elephant and owl, which they want to enjoy the music too.
This is also related: https://people.xiph.org/~xiphmont/demo/neil-young.html
Samplerate list (SEO)
1000 Hz
2000 Hz
3000 Hz
4000 Hz
5000 Hz
6000 Hz
8000 Hz
8363 Hz - Amiga low C
10000 Hz
11025 Hz - Soundblaster
12000 Hz
16000 Hz
16726 Hz - Amiga high C
20000 Hz
22050 Hz - SoundBlaster Pro
24000 Hz
32000 Hz
33075 Hz
40000 Hz
44100 Hz - CD
48000 Hz - DAT
64000 Hz
66150 Hz
88200 Hz
96000 Hz - Hires audio
128000 Hz
176400 Hz
192000 Hz - Artifact study audio
256000 Hz
352800 Hz
384000 Hz - Dolphin audio
512000 Hz
705600 Hz - lowest common multiple of 44100 and 48000, used for resampling
etc. - 32000*2*2*2*2*2..., 48000*2*2*2*2*2..., 44100*2*2*2*2*2...
2116800000 Hz - 44100*48000, divided by 2 until it can't gives 4134375 Hz.
14112000000 Hz - lowest common multiple of 44100, 48000, and 32000. Divided by 2 until it can't gives 6890625 Hz.
From Facebook:
kHz: 8, 11, 12, 16, 22, 24, 32, (33, 40,) *44*, 48, (64, 66,) 88, 96, (128,) 176, 192, (256, 353, 384)
bps: 8, 10, 12, 14, *16*, 20, 24, 32, (64, a-law, u-law, 16F, 32F, 64F)
chn: 1: 1.0, 1.1; *2: 2.0, 2.1, 3.0, 3.1*; 4: 4.0, 4.1; (6: 5.1, 6.0, 6.1; 7.1, 22.2)
mode: PCM, (delta, deltabit,) DSD, unsigned, FLAC, (WV, WMAL, ALAC, ZIP, RAR, 7Z)
6.0 a 6.1 - each speaker on 1 side of a cube. 3.1 can be transmitted as stereo with 1 center channel playing both.
No comments:
Post a Comment
Barely anyone comments, so I don't moderate. Free advertising, I guess.