[QUOTE=tubebar;2557266]Downsampling a freq is going to lose quality (and lose the freq), you always lose when you go down.
When changing to lower frequency, you are losing your higher frequencies, and then estimating the remaining frequencies.
[QUOTE=tubebar;2557266]Again it shouldn’t make any diff in going up. Yes if you go back down you will lose quality compared to the original won’t be 1:1 if that’s what you’re saying?
You will always lose quality, because you will always be [U][I][B]estimating[/B][/I][/U] the value of the new sample depending on the time when the [I]new [/I]sample falls between the two [I][B]REAL [/B][/I]samples, and the value that the new sample takes will be a sample of a RECONSTRUCTED waveform.
The new sample will be from a reconstructed (ie; estimated) waveform, and is, therefore, not accurate.
While waveform reconstruction algorithmns these days are pretty good, and you are unlikely to hear the difference, you have still lost accuracy.
In 5 years time, the reconstruction algorithmns may be using increased higher orders, which provides better (more accurate) reconstruction … however, you’ve already resampled your original waveform with an older algorithmn with a lower order, and probably deleted the original as well … therefore, you’re forever stuck with an estimate of the original recorded waveform.
In the event that you resample to a multiple frequency (lets assume 2x), the algorithm spits out a new sample between the two REAL samples, which is the closest representable sample of the reconstructed waveform between the two REAL samples [I]based on the algorithms used today[/I].
When you reconstruct the waveform (for playback), the player is interpolating (ie; estimating) values between a REAL value, and an interpolated value … which is less accurate then interpolating between two real values.
Again, if reconstruction algorithms are improved between now and a few years in the future and you’ve resampled your original recording to a higher frequency sample rate … you’re stuck with the crappy estimate.
10 years ago, most wave reconstruction was basically linear on cd players, and premium players reconstructed waveform based on squared reconstruction (2nd power), whereas now, it’s based on a cubed reconstruction (3rd power)… and on that it could be even higher power filters. I’m not entirely sure, because I haven’t been following CD players recently, and have neglected my studies on where audio processing software has gone in the last 5 years
As technology progresses, processing power increases and hence sound reproduction quality increases …
[QUOTE=tubebar;2557266]You completely lost me when talking about years? How would time change a data file? Aside from errors that can be avoided w/ proper archiving.[/QUOTE]
Resampling to different frequencies = loss of quality, and possibly accidental removal of higher (audible) frequencies.
Resampling to lower bit depth = HUGE loss of quality.
This discussion excludes the accuracy of storage devices.