Phil Leigh;578522 Wrote:
> OH - and I thought you were about to understand! :-)
>
> 1) all 16/24 bits represent the quiet sound, same as they do a loud
> sound. The fact that several of them are zero makes no difference
> except to the SNR. You have to stop thinking about "bits" as things in
> their own right. They aren't - they are just part of a "word" that
> represents the level (loudness) of something when you add ALL of the
> bits together.
>
> 2) You also have to leave behind the digital photography analogies
> because they aren't helping you to understand at all. In photography
> each pixel is equivalent to a sample and has a bit-depth. BUT and it is
> a MASSIVE "But"... when you look at a digital image you see all the
> "samples" at once fixed in time, whereas in audio the sound is produced
> by a variation in amplitude over time - that's all sound is! So in
> audio, it is the change over time that matters. This is why pictures
> get "blocky" when you zoom in - eventually you can see individual
> samples - which make no visual sense in isolation.
>
> There is NO equivalent to this in audio - without special tools (DSP)
> you can't alter the timeframe in which you hear the sounds - and the
> sounds themselves change (in pitch and duration) if you try. You can't
> hear a single audio sample - in isolation it has no "sound". You can
> see a pixel.
>
> 3) I don't think you've grasped the meaning of precision vs resolution
> yet. In your 7-bit example, there are NOT only 7-bits worth of values
> to represent the sound... 7-bits is just what it takes to represent the
> LOUDNESS of THAT sound. The fact that the higher bits have zeroes in
> them doesn't mean anything for the quality of the sound other than it
> is quiet so you have more chance of hearing the noise floor intrude
> into it.
>
> If the quiet sound was made louder it would be captured with the same
> precision, but a greater SNR.
>
> {upper case for emphasis]
> IN LINEAR PCM, RESOLUTION IS A FIXED PRODUCT OF BIT-DEPTH. It does not
> vary with volume.
> In lossy compression (MP3 et al) bit-depth...and thus
> resolution...varies.
>
> 4) Your optical vs digital zoom analogy is not provocative - merely
> wrong :-)
> Analogue sources do NOT have infinite resolution. Nor do our ears, in
> fact - but that's way OT.
>
> Tape is limited by several things, including the tape speed, tape
> width, tape head gap and magnetic particle size, all of which culminate
> in a finite resolution. Likewise, vinyl is ultimately limited by the
> physics of both the cutter head, the replay stylus, the pressing
> process and the material properties of the final vinyl - all of which
> limit the accurate recording of tiny (quiet) details.
>
> I'll casually ignore the fact that most vinyl is cut from digital
> masters in the last 20+ years... :-)
>
>
> Replaying quiet sounds is NOT the same as recording with less bits!
>
> Try the SB volume control - as you wind the volume all the way down the
> sound doesn't get nasty & grittier like a pixellating image - it gets:
> 1) quieter
> 2) noisier
> 3) eventually some of the quietest parts will no longer be audible...
> just like they would with any analogue source... this is simply because
> they are too quiet to be heard, both in relation to the noise floor and
> to the loudest parts.
I have been studying this quite carefully as there are lots of useful
points. I promise it wasn't point 1 I was missing and I don't think it
was point 3 although I guess I do find the terms resolution and
precision confusing. (I have been puzzling over the question of what
the precision of an analogue system is)
I also undertake never to allow my mind to wander into visual
analogies.
What I am getting confused about is what happens when a sound (which,
obviously as you point out, has an amplitude that varies over time)gets
louder and softer. Let's assume there is no other sound in the
recording. If the sound (is it allowed to be a sine wave?) varies in
amplitude between 128 and 32- xxx...10000000 and xxxx....00100000
and it became louder so that the peak amplitude was 2048
xxx....100000000000, what would be the number which now represents the
minimum amplitude of this wave? Would it be 1952 (11110100000 ?) or
something else?
I have a feeling that if I could grasp this I could understand the
point.
--
adamdea
------------------------------------------------------------------------
adamdea's Profile: http://forums.slimdevices.com/member.php?userid=37603
View this thread: http://forums.slimdevices.com/showthread.php?t=82050
_______________________________________________
audiophiles mailing list
audiophiles@lists.slimdevices.com
http://lists.slimdevices.com/mailman/listinfo/audiophiles
