Originally Posted by
Megaohmz 
Remastering in 24 bit vs. 16 bit is a no brainer. You get better results with the 24 bit file than with the 16 bit. Imagine a sine wave (~), the "bits" are used to represent the steps used to map out the distance going up and down on the waveform itself. The more bits you have the finer the steps become and the more resolution you will have. It also dependes on the sampling frequency of the 24 bit file as well. If you have a 96 KHz sampling frequency this will be better than twice as fine as a CD which has 44.1 KHz sample rate. The sample rate is the criteria for how wide the steps on the sine wave will be. The DAC (digital to analog converter) will then take the numbers and turn it into a sound that is pretty close to the original. The finer the steps, the closer to the original it will be.
The reconstruction filters get rid of the "steps", it's a wholly inaccurate image of what's really happening. Whatever sample rate you are using the filtered analog signal has no "steps". Also, resolution is a matter of sampling rate, dynamic rage is a matter of bit depth, both notions are different. In general, it's true that the greater the sampling rate/but depth is, the more accurately you can digitize a signal, but in audio, you first run into the limits of human hearing (20000 Hz and 0 dB SPL at its most sensitive range) and as important, the lack of musical content above 20000 hertz and the ambient noise of your listening room.
Again the bits are the (rise) of the "stairs" and the sample rate is the (run) (if you are into linear algebra)! I hope that I explained it well. Actually it is very common to see 24/196 in all recording studios now, since they do alot of mastering there, it makes sense to use those rates, even if your endgame is the "lowly" CD, or a dreaded MP3 file. 5,644,800 Hz or 5.6448 MHz sampling rates are possible and used today.
It makes sense to use them in studios because they are doing a lot of editing and thus limiting noise and calculations artifacts is a huge bonus, it makes much much less sense for playback. Besides, comparing DSD sample rates to PCM sample rates is confusing to the reader since it's a totally different technology, most of the DSD coding is noise, it relies on pulse density modulation and the frequency used has a totally different purpose unrelated to to the Nyquist frequency of PCM (pulse code modulation).
If you try to sample a real world sound from a band playing hundreds of frequencies at a time and only use 44.1 KHz sampling freq, then you can draw a pretty good 22 KHz sine wave. But since there are a bunch of other frequencies on top of the others it makes more sense to use a faster sampling rate to get the most resolution possible, and try to map out the complexities of all of the combined sounds. That's why you hear so much more detail in the higher bit/ higher sample music.
Again you fail to understand the basis of a linear system and the frequency domain, adding soundwaves of different frequencies don't make you suddenly need higher frequencies to represent them. Unless you somehow postulate that higher than 22000 Hz frequencies are necessary for reproducing music even when they are inaudible, there's no reason to encode higher than 22000 Hz information.
With the Frequency response of the headphones it is a rule of thumb (sort of) that if the freq response is wide like 15Hz-28KHz you have a better quality driver in the phones. But that is not what makes them sound good, for the most part. It just means that the phones are capable of producing dB (sound pressure, decibel) at 15Hz as well as 28KHz and everything in between. The dB will change dramatically between the frequencies and a good phone will be as flat (not change dB) along the frequencies. But all headphones are different and none are perfect.
Agreed on that part, the frequency range given is in no way indicative of the final performance of headphones.
It is actually theoretically impossible to make a totally flat responding speaker, because of the change in impedence with the changes in frequency. The more impedence the coil has the less dB you get. The higher the frequency the more impedence you get in an inductor, which is what makes the speakers move less, giving you less dB. There are other speaker designs that don't use coils for motors, and they are more expensive, and have different issues with physics.
It's quite possible to make a totally flat (with 2 dB in anechoic conditions) speaker, a number of manufacturers certainly succeeded.
