Quote:
Originally Posted by Crowbar /img/forum/go_quote.gif
You're way off base here.
There are two ways beat frequencies can be heard. It's known that the brain will form beat frequencies when you play one frequency in one ear and another in the other one. But if the frequencies are ultrasonic, that doesn't happen since they're not detected in the first place. Now, with ultrasound it's possible to create and audible beat frequency, and that's because of the way sound intermodulates in the ear. However, the same applies to bandlimited recording hardware (and all of it is). ADCs used for recording with the sort of filters they have would cause the same intermodulation to be present as actual 5K signals. During playback, that will play back simply because the intermodulation that your ear normally does would have happened at the microphone and ADC and was recorded as a real frequency.
|
There has been a lot of discussion about this, and as far as I know the conclusion simply hasn't been reached yet. If a certain system (or room!) has an imperfect transfer function, then these difference frequencies occur, and indeed a 50K and a 60K 'source' sound will result in a 10K target sound. The question is wether or not recording gear should and will record this resultant frequency, or should (also) store the two source frequencies that created it.
Quote:
| Oh wow dude, epic failure in understanding basic sampling theory here! Shame... Nyquist limit's minimum 2x sampling frequency is exactly for sine waves; square waves are not bandlimited as it takes infinite sampling frequency to reproduce a true square wave, unless you change from a Fourier basis to say Haar wavelets (and not possible in the physical world since it requres infinite slew rate). You couldn't have possibly gotten this more backwards. Ouch. |
Yeah I got it twisted around, my bad. The argument still holds though, as you graciously confirm.
- If you need 'unlimited' sampling frequency to reproduce a true square wave, then the higher samplefrequency reproduces a square wave better.
Quote:
| Wrong again. Digital and analog are just different ways to represent information. The 20 kHz bandlimit of human hearing directly translates to 40 kilosamples/s to encode the same information digitally, which needs to be raised for practical limits (space for the filter). Likewise for dynamic range and bit depth (20 bits is sufficient in optimal system = anechoic chamber). More generally, the discreteness limiting precision of digital systems doesn't make them inferior, since the physical world is also of limited precision (proof by contradiction: infinite precision analog implies infinite information density, which violates the Bekenstein bound). |
That is simply one way of looking at things. I hold that the 'perfect' digital audio would be at infinite bits, and infinite khz. Naturally there are all sorts of more or less subtle differences, but this point I was making wasn't intended to be the end-all of all arguments.
Quote:
| Incredibly limited dynamic range and distorted frequency response requiring RIAA equalization. Oh please. |
Well only cause you asked so nicely.
Quote:
Originally Posted by Crowbar /img/forum/go_quote.gif
The filter issue is taken care of by oversampling. I think the real reason for 192's existence is "more is better" type marketing.
|
For $100 DACs, Im sure that might be the case. But specifically with the target audience (that is definately not impressed with things like 89578923475234 PMPO WATT POWER!@#) I dont think that hardware designers in the (semi) audiophile segment will be -that- influenced by the pure '192 is larger than 96 so it must be better!' argument.
My personal reasoning for considering 192K:
- 192 is the limit of DVD-A, and seems to at least be a number quite often used to indicate the top end of the spectrum. Leaving aside for a moment wether or not DACs can do it -now-, I do not think that in the coming 10 years (and perhaps even our lifetimes!) there will be a need to up bitdepth and samplerate, simply cause the sources don't even hypothetically support it yet. (even BluRay and HD-DVD 'only' specify a max of 192/24
- High end recording outfits commonly record single sources ('microphones') at 96, after which one would sensibly upconvert to 192 inside the computer to start doing the 'DSP work' i.e. mixing/mastering, and this 192-mastered stereo track has only relevant data. It's not just upconverting-for-upconverting's sake.
- The '192' manufacturers I am looking at do (of course!) claim that their gear does work satisfactorily at 192. While they need to live off their work and might even lie about their choices after mass production started, the company I chose have been working with custom digital gear since 1999, so at least they probably won't make beginners mistakes. It would be interesting to invite someone from that company in here to debate the issue with Benchmark, but I suppose this topic is 'Benchmarks' and competition would not be welcome (at least not in this specific topic)
