You've seen mine analysis of the presentation? Maybe you can explain how Watt's theory of audible but immeasurable noise floor modulation and crazy low noise shaping resolution works, since you seem to support it?
I like his idea of creating extra samples to reconstruct higher resolutions based on the sinc function
and I will repeat my reaction to j-j here.
"Concerning my thoughts on the "sinc issue";
From the
theoretical side:
Regenerating the perfect analogue reconstruction of a sampled band limited signal is
only possible by the sinc filter function.
This also means we should take all samples of a music piece into account (thus define a begin and an end)
It also means it will take eternity before we can replay the music piece as the sinc impulse response is not time limited
Conclusion; perfect reconstruction is not possible as delay on playback cannot be unlimited
Conclusion: perfect recontruction is not possible as it is impossible to realize a delta T = 0 between samples
From the
practical side:
compromise to limit the amount of generated samples between the CD samples (15 chosen in the subject)
this will limit the accuracy of higher frequencies mostly
compromise to limit the respons time of the filter
this will limit the accuracy of lower frequencies mostly
In case compromise the respons time window of the filter to one second:
the lowest contribution wil be the first and last sample of the sinc window is sin(jwt)/jwt and 1/(44.100Hz * 0,5 sec) = let's say - 88 dB maximum and although not major, might still be significant
Contributions of samples more in the middle of the timeframe of course might have larger contributions to the extra generated samples.
Anyway; Limiting the amount of extra calculated samples between any two original and/or limiting the timeframe used for this calculation of the extra samples is throwing away information that is all available in the original sample pattern
This lost information can never be regenerated by any means once it has been thrown away.
Of course the main question remains;
At what compromise is missing information turning out to be audible ? (for someone/anyone)"
Please elaborate on what is fundamentally wrong with the above.