Message Sort: Post Order or Asylum Reverse Threaded
horizontal resolution
24.242.83.220 |
||
Posted on April 21, 2015 at 11:29:06 | ||
Still trying to have a better understanding of digital audio. My ears tell me that higher resolution formats sound better. At the same time, I am a little mystified by the opposition being expressed at the movement beyond redbook CD. I bumped into an article that was interesting and found a concept with which I could identify and was very well articulated (see below). Thanks Joe!: Joe Baldwin, Virginia Beach, USA With all do respect to the experienced engineers and sound technicians I must disagree. This disagreement is on a fundamental level and with the premise of the original article. The "facts" listed are not all of the facts and do not do justice to the bigger issue, (no matter how big a font the word "fact" is presented). If you stay with me a bit, I think that you might agree. I have friends that claim to be audiophiles, but are really techno-weenies with too much money, and I have other friends who can actually discern differences in audio component in double blind tests simply because they have extraordinary hearing as well as technical knowledge. To the latter group, CD's (Redbook 44.1KHz/16bit) are clearly not as true a medium as high quality vinyl. There is a good technical reason for it. It appears to be resolution, encoding/decoding errors, and encoding/decoding standards errors. Repairing the standards errors, advent of higher quality A/D D/A converters (no matter the resolution) has gone a long way to improve the digital medium. With respect to resolution: it appears that many recording professionals have warmed up to an increase is resolution. Lets be specific here: the vertical axis representing amplitude has been steadily increasing from 8 to 16 then to 24 bits, and now some software has standardized on 32 bit internal processing. The horizontal axis representing time has increased (recently) from 44.1 samples per second to 48, then 96, and on up to 192. It appears from the current discussion that increasing resolution in the vertical axis is assumed to be G.O.O.D while increasing resolution in the horizontal direction is assumed to be B.A.D. The only real cogent reason (other than misunderstanding) for this is that increased bit depth is much easier to store and process with current computer technology than is sample rate. High sample rates mean costlier computers and M.U.C.H bigger hard drives, and S.L.O.W software. So why is the horizontal axis resolution increase not invited to the party and the vertical axis resolution embraced? Good question. Lets table that question and get to the F.A.C.T.S. Here is a table of Horizontal Resolution: 48 KHz sample rate Table Audio Frequency | Number of Samples/wave ------------------------------------------------ 24 KHz | 2 samples/wave 12 KHz | 4 samples/wave 6 KHz | 8 samples/wave 3 KHz | 16 samples/wave 1500 Hz | 32 samples/wave 750 Hz | 64 samples/wave .... and so on Kind of surprising F.A.C.T.S when lower frequencies are considered, isn't it? Therefore, this is not just a discussion about super high frequencies, this is ultimately a discussion of Horizontal axis waveform resolution. Now for Vertical resolution: Bit-Depth Table # Bits | Discrete Amplitude steps ---------------------------------------- 8 | 256 steps 16 | 65,536 steps 24 | 16,777,216 steps OK, so far we have 16 Million steps in the vertical direction vs 'way not many' in the horizontal. Even the most entrenched Horizontal-High-Resolution-phobic among us will observe that even in substantially mid audio frequencies such as 1500 Hz while there is over 16 Mega-Steps in the Vertical there are only 32 Steps in the horizontal. Can your ears hear it? Yep. Do we really need HD TV at 1080 Horizontal scan lines when we have Super Video at around 400 scan lines, or 16 Million colors? Yep. In fact I would argue that if the technology could support it, 192 KHz sample rate is not nearly enough. Crazy? Well consider this factiod: at 192 KHz sample rate we would be up to a whopping 256 samples for a 750 Hz wave. Dude, that is like almost 8 bits rotated to the Horizontal axis for a substantially well-within-the-hearing-range of non-audiophiles and old people :). Finally, check out this (very daunting) analysis of "Nyquist Theorem's Consequences": http://www.digital-recordings.com/publ/pubneq.html#theorem It appears to confirm my 3rd grade math-tool analysis above (as well as my ears), that the trouble with adhering to the Nyquist Theorem in a pedantic manner results is more errors as the frequency increases (starting at relatively mid-range frequencies). Said simpler: as the frequency goes up so do the errors and thus distortion as a direct result of decreased resolution per wave. My conclusion is that given what my hearing tells me, the low resolution of samples/waves, and the Nyquist Consequences analysis, current sample rates per wave must be improved upon in order to faithfully capture mid-high frequencies that are demanded by audiophiles and others. Personally I think, when we get to the point of having on the order of >= 65,000 (or even 1/10 that number) samples per wave at 15Hz-10KHz-audio, then we can start a reasonable debate about what is 'too much' resolution in the Horizontal direction. Until then, striving to increase the Horizontal resolution while not overburdening a CPU (or the pocket book) is an exceedingly worthwhile endeavor for the technical recording artist. Wednesday January 18, 2006 |
RE: horizontal resolution, posted on April 21, 2015 at 17:43:00 | |
Posts: 15524
Location: Alabama Joined: September 11, 2010 |
Thanks! for sharing- wolfy. Really interesting info. |