24bit vs 16bit, the myth exploded! - Page 16

You've slightly misunderstood. If the sample rate is at least double the audio frequency, the theory states that the waveform can be reproduced perfectly. The important word here is "perfectly", not as you have stated above, "possible". The term "Perfectly" is used here in it's literal, superlative sense and means in this context a completely linear output, IE., with no distortions or artefacts of any kind and containing 100% of the original waveform and ALL it's detail. Using higher sample rates cannot make the reproduced waveform any more detailed or higher quality or more perfect.

If you have a quick read of my last post again, perhaps it will make more sense now you know what I mean by "perfectly".

What Publius said in his reply to you was correct.

It's possible that resampling may cause distortions but generally speaking I wouldn't expect there to be anything too noticeable. Phase differences are very unlikely to be introduced. To hear the effects of phase differences usually requires differences in the milliseconds range (1/1000 sec or more). A DAC should easily be sample accurate at 44.1kFs/s. The duration of a sample at this rate is 1/44,100 sec, far more accurate than the few milliseconds required. If you can hear some phase issues when oversampling there is something fairly seriously wrong with the DAC.

You would need to be a bit more specific. If you mean during playback of a mixed master at normal hearing levels, then no I hear no differences. If you mean while mixing a channel of audio with a poor SNR and substantially boosting, say by 60dB or more (1,000+ times), so it will balance when other channels are added to the mix, then yes I have heard a difference. This is of course the whole point of higher than 16bit for recording. Be careful not to get hung up on 32bit. There are some difficulties when using 32bit due to the way an audio software program has to deal with exactly where the decimal point is. In other words a direct comparison between 32bit and 16bit may or may not provide the expected additional 16bits of dynamic range. This is in theory of course, in practice, as previously mentioned, the practicalities of electronic circuits make the full dynamic range unobtainable anyway. If you are working professionally, 24bit in practice probably provides slightly more dynamic range (through a lower noise floor) than a 32bit system.

G

The answer to that is a qualified yes! By this I mean that releases in 24/96 sometimes sound better than at 16/44.1. Not because there is any real advantage to the format but simply because more effort is often put into the mixing and mastering process. To get round this what you could do is own a 16/44.1 DAC, download 24/96 recordings and convert them to 16/44.1 for playback.

G

For an even more accurate understanding of digital filters have a read of this article: http://www.users.qwest.net/~volt42/c...ng/Filters.pdf

As I mentioned earlier in this thread, there are various methods in use at the moment for reconstructing the digital audio. Oversampling works well for some DAC manufacturers but better implimentation of 44.1kFs/s filters works better for others. I'm not sure if I agree with your analogy, for it to be accurate, every word would not only need to be repeated twice but would also need to be repeated twice as fast. If anything, this additional overhead is likely to cause more errors rather than fewer. I'm not saying that oversampling DACs are worse in practice, I'm just saying that there is no such thing as perfection in practice.

I have to say that the reconstruction technology in modern good quality DACs has probably already reached the point where the analogue circuitry in the DAC after the conversion has more effect on sound quality than the actual conversion process itself. If this statement isn't true at this instant, it's getting closer to being true all the time. In other words, what we are talking about here, with regard to sample rate is going to be inaudible to the vast majority of consumers, what happens in the mixing and mastering is going to have a far more obvious impact on sound quality. I personally think there is more to hi-res audio than pure placebo but I also believe that the vast majority of the difference between hi-res and CD can be accounted for by the mixing and mastering process rather than by the differences between the digital formats.

I should point out that I do not have an indepth understanding of the practical application of anti-alias and anti-image filters. With this in mind, it is my opinion that at this point in time, an 88.2k or 96k sample frequency may sound marginally better on some DACs. For this reason I believe that the best digital audio format at this point in time for the consumer is 16bit 88.2kFs/s. Strange that no one seems to make this format available to consumers.

To be honest I'm not very familiar with consumer software for this process. There is an organisation which makes a dither called Pow-R. You might want to do a search and see which software includes this dither. Pow-R is good quality dither and is administered by committee, so it might be that there is source code available or at least a list of software which includes it.

G