Originally Posted by cnocbui
Well if a microphones can't, then it cant be recoded in the first place so you don't need the capability in the reproduction system
I know about the extent of the dynamic range of human hearing. 16 bits gives you a dynamic range of 96db What is the signal to noise ratio of a decent power amp? About 100 db. So what's the point of having 24 bits, giving a theoretical dynamic range of 144 db? You can't reproduce a sound that is quieter than the noise floor of your system. With nothing playing on my system I can hear a faint hiss, it can't go quieter than that.
The average home has a background noise level of about 50 db, the ambient noise floor if you like. So 50+96 gives 136 db peak volume while still allowing the quietest sound in the recording to be reproduced. But that is 16db over the pain threshold of 120db, so 96 db is plenty. If you had a listening room that was constructed like a professional sound recording studio with a background noise level of 20 db, 96db is still good enough for a peak spl of 116 db. How loud is that? between the sound of a pneumatic riveter at 1m and a chainsaw at 1m.
16 bits is enough, 24 is beyond overkill.
The brains ability to filter is a function of a signal processing ability and is a red herring with regard to bit depth and sampling frequency.
The dynamic range from ordinary CDs is limited. The range comes in two versions, one is the absolute range, that is the energy difference between the smallest signal and the largest. The smallest signal level is often the noise level, that is, the sound you get just before a music track starts. The absolute dynamic range is therefore in effect equivalent to the signal-to-noise ratio. The other one is the relative range, that is the difference between the largest and the average energy. This is probably more important, as some records are engineered to make as much volume as possible because hat will make them heard more in public places than average tunes and thus more succesful in the marketplace. To get this effect, the volume is pumped up. As a result, the maxima (the spikes in the music volume) have not enough power left to be played fully. As a result the sound is loud, but compressed. The sound is `dull' and `dead' if there is unsuf cient dynamic range to create a true playback. The small relative range means that the full theoretical range of a medium is simply not used, so we concentrate here on optimal recordings and thus the absolute dynamic range, the difference between maximum signal power and minimum signal power.
It is common knowledge that CDs have a lower noise level than vinyl LPs. But the situation is not that bad for vinyl. In reality, noise from a vinyl record is concentrated in the lower frequencies (say, below 500Hz) and in the rest of the spectrum, the noise levels of vinyl analog LPs are comparable to CDs.
But the story does not end here. Because, surprisingly for many, the ear functions far better than our best technology. Unlike microphones, the ear is able to detect signals of a size far less than the noise level, signals with a power of 20-25dB below noise level. 25dB equals 2.5B, equals 10 to the power of −2.5, hence -25dB less power means 316 times less power. As signal power is related to the square of the signal level, 25dB less therefore means a signal that is 17 times smaller than the noise it is embedded in, is detectable by the human ear. The ear is capable of this feat because - as research has shown - it is in effect not one single sound measurement device, but thousands and thousands in parallel, all tuned to a specific frequency and all with active reduction of noise. That has two possible effects on experienced sound quality.
One is that the first 20dB (90%) of noise are far less relevant. They do not obstruct listening to the sound. They might bother you psychologically, but most people are able to mentally switch them off, in the same way they can switch off the noise of a crowded and noisy bar when talking to someone. We listen not just with our ears, but with our brains as well and our brains are champions of finding relevance out of noise. Hence, it is not a red herring, as it allows us to ignore noise that is far louder than the actual signal. And that is the essence, we can listen inside
noise, so noise levels are less relevant than they seem.
The other is that you will experience more distortions generated by the DAC-process. After all, the smaller the signal, the less level-steps are available for the DAC to turn the result into the smooth harmonics nice sound has. The theoretical highest signal level on a CD is 65536 times (16 bits, or 2^16) higher than the lowest. That results in 65536 × 65536 (or 2^32) in terms of signal power or 10 × log (2^32 ) or 222dB. But the DAC needs a certain minimum of level-steps to be able to produce smooth analog results out of discrete digital steps. Suppose it requires at least 9 levels to produce a fair smooth analog signal, the lowest signal level is 9 bits and the highest level is still 16 bits. The actual theoretical dynamic range is then indeed around 97dB.
URL on the human ear:http://www.princeton.edu/~wbialek/ou.../bialek_87.pdf
I was wrong saying that microphones cannot register this, they can.