Digital Storage Format Questions (updated tl;dr (yes!) and removed OP for new, shorter version)

I clarified because someone asked me to shorten. I did shorten it, and now I've replaced the original post with it. It's less detailed but there was too much and, probably, some [more?] potential errors.


NEW POST:
DSD64 is the same bitrate as 16/44.1, yet sounds better, close to 24/44.1

The reason DSD sounds better than 16/44.1 at the same rate, is, when vertical resolution (dynamic range) isn't necessary (the current level of the song is at a lower than maximum level), the extra storage space goes to storing detail in the horizontal (time) aspect - effectively increasing the sampling rate and decreasing the bit depth. But the bit depth isn't decreased.. It is, but only what's not being used.

DSD can carry much higher frequencies that 16/44.1, MUCH higher. Amplitude can be so low that over 1,000,000Hz can be stored! (clarifying: amplitude must decrease for higher frequencies and significantly for such extremely high frequencies!) (recording 1Mhz it's useless for obvious reasons - its below the DAC's output capability, above the DAC's low pass filter, below hearing threshold even if properly rendered). The reason I stated this uselessly high frequency is to show the method by which the sample rate is effectively increased in DSD64 and the difference between DSD64and PCM 16/44.1

-DSD64, uses all of whatever's not used for vertical resolution (amplitude) for horizontal (time/phase accuracy). 16/44.1 has about 50,000 horizontal and 50,000 vertical points to plot the waveform per second. This gives a dynamic range of about 96dB and the ability to store frequencies to about 22kHz.
-DSD islinear like PCM and it has a ~20dB higher max dynamic range. We know, if loud transients are required, DSD is able to deliver. Usually extra-loud parts in a song (+115dB?!) are events numbered in the single digits, not kick drum. Since these extremes are so loud and infrequent, their time accuracy down to the nanosecond isn't critical so much as maybe the upper harmonics of a few instruments with very textured and layered aspects to their character at a level down around 70-80dB (not ear-splitting)

What I want to do, is figure out around how much DSD64 expands the effective sample rate during these 70-80dB passages which people commonly believe DSD64 to sound better than 16/44.1

Why? Because then, We could improve from 24/44.1 resolution, even though we're already recording at 24 bit and covering 20Hz-20kHz without HF roll-off. How?

- Since DSD64 (2.82Mbit/s) improves on quality of 16/44.1 (2.82Mbit/s) while using the same raw storage space without compression by dynamically trading currently excessive dynamic range capability for horizontal resolution (time/sampling) whenever extra resolution is available at the constant bitrate, all of the sound quality improvement noticed on DSD64 during time where the level of the DSD recording is at or below about -20dB is due to he extra (effective) sampling rate.

The improvement from going 16/44.1 to DSD64? I want that to happen again going from PCM 24/44.1 to 24/xxx, say 384, tentatively. I can't make it clearer without making this longer - you get the aim, right?

Just trying to figure out if it works the way I think it does and then, maybe, eventually, this understanding can be used to estimate the sample rate one should use during recording for an audible improvement over 24/44.1 at 24/xxx, a difference, hopefully, somewhere around the difference between 16/44.1 and DSD64!

(this is useless for resampling - the data simply isn't there unless you're doing something that's already been recorded at DSD1024, and then why would you want to degrade it?! This is for showing that it's not just a waste of space t choose 24/44.1 over 24/xxx (say 384) so long as the increased rate begins during recording.

 

Need a TL;DR on this one. Are you simply saying higher sample rate DSD should sound better than higher bit rate PCM? Plenty of people do software upsampling on DSD because they believe this is so.

 

DSD64 is the same bitrate as 16/44.1, yet sounds better, close to 24/44.1

The reason DSD sounds better than 16/44.1 at the same rate, is, when vertical resolution (dynamic range) isn't necessary (the current level of the song is at a lower than maximum level), the extra storage space goes to storing detail in the horizontal (time) aspect - effectively increasing the sampling rate and decreasing the bit depth. But the bit depth isn't decreased.. It is, but only what's not being used.

DSD can carry much higher frequencies that 16/44.1, MUCH higher. Amplitude can be so low that over 1,000,000Hz can be stored! (clarifying: amplitude must decrease for higher frequencies and significantly for such extremely high frequencies!) (recording 1Mhz it's useless for obvious reasons - its below the DAC's output capability, above the DAC's low pass filter, below hearing threshold even if properly rendered). The reason I stated this uselessly high frequency is to show the method by which the sample rate is effectively increased in DSD64 and the difference between DSD64and PCM 16/44.1

-DSD64, uses all of whatever's not used for vertical resolution (amplitude) for horizontal (time/phase accuracy). 16/44.1 has about 50,000 horizontal and 50,000 vertical points to plot the waveform per second. This gives a dynamic range of about 96dB and the ability to store frequencies to about 22kHz.
-DSD islinear like PCM and it has a ~20dB higher max dynamic range. We know, if loud transients are required, DSD is able to deliver. Usually extra-loud parts in a song (+115dB?!) are events numbered in the single digits, not kick drum. Since these extremes are so loud and infrequent, their time accuracy down to the nanosecond isn't critical so much as maybe the upper harmonics of a few instruments with very textured and layered aspects to their character at a level down around 70-80dB (not ear-splitting)

What I want to do, is figure out around how much DSD64 expands the effective sample rate during these 70-80dB passages which people commonly believe DSD64 to sound better than 16/44.1

Why? Because then, We could improve from 24/44.1 resolution, even though we're already recording at 24 bit and covering 20Hz-20kHz without HF roll-off. How?

- Since DSD64 (2.82Mbit/s) improves on quality of 16/44.1 (2.82Mbit/s) while using the same raw storage space without compression by dynamically trading currently excessive dynamic range capability for horizontal resolution (time/sampling) whenever extra resolution is available at the constant bitrate, all of the sound quality improvement noticed on DSD64 during time where the level of the DSD recording is at or below about -20dB is due to he extra (effective) sampling rate.

The improvement from going 16/44.1 to DSD64? I want that to happen again going from PCM 24/44.1 to 24/xxx, say 384, tentatively. I can't make it clearer without making this longer - you get the aim, right?

Just trying to figure out if it works the way I think it does and then, maybe, eventually, this understanding can be used to estimate the sample rate one should use during recording for an audible improvement over 24/44.1 at 24/xxx, a difference, hopefully, somewhere around the difference between 16/44.1 and DSD64!

(this is useless for resampling - the data simply isn't there unless you're doing something that's already been recorded at DSD1024, and then why would you want to degrade it?! This is for showing that it's not just a waste of space t choose 24/44.1 over 24/xxx (say 384) so long as the increased rate begins during recording.