Tuesday, October 16, 2012

Digitizing The "Analogacal"

The technological advances behind recording audio never ceases to amaze me.  Whether it was Edison and his wax cylinders, Berlinger with his original flat record concept, or even the digital era, just think how amazing it is to preserve moments in time forever, to be shared by generations.

I pretty much took all of it for granted.  However, recently I've re-disk-covered (sorry for the pun) my LP collection.  Still unbelievable that my 40+ year-old records still sound pretty good, despite some scratches and such along the way.

Then the CD era came in.  Wow, all that LP on one little disc that fit in your hand.  No needle ever touches this thing.....it will last forever.  how did they do it?  All this talk about pits and stuff, and how the laser could interpret those pits and determine the analog equivalent.  Huh? 

I'm not going to get into the math or physics in any great depth here.  If you're looking for that detail, I welcome you to google your ass off.  There is a TON of info out there.  You just have to go look for it.  It's there.  I'm going to talk in my own language and hope it makes sense.


The sampling rate and the bitrate are perhaps the two most important things to remember when discussing how an analog signal (or sine wave) gets converted to digital.

SAMPLING RATE is simply the number of times an analog signal is sampled in ONE SECOND.  For red book standard CDs, the talk is always 44,100 hZ.  That means that the analog piece is actually sampled 44,100 times PER SECOND!!

Why 44,100 hz?  Well, there is some extremely complicated math (Nystrom theorem?) that in a nutshell says that you must sample at twice the rate of the frequency you are trying to capture.  Remember that the red book standard says that only frequencies in the 20-20000hz range will be captured on a CD, so 2 x 20000hz = 40,000 hZ (so 44,100 hz sampling rate pretty much covers that.

The next step is what do we do with each sample?  We have 44,100 of them now.  Each sample at that given point in time on the analog wave will be assigned a binary number.  Remember the binary number system?  You can never have a number bigger than 1.  So 0=0, 1=1, 2=10, 3=11 and so on.  This is also called the bit rate.  Now imagine that EACH SAMPLE is assigned a 16-bitrate (2 to the 16th power).  That means that each sample is capable of being assigned a number that looks like

                                     xxxxxxxxxxxxxxxx  where the x could be either 1 or 0. 

That also means that EACH SAMPLE has a combination of 65,546 "levels" that can be assigned to it.  One sample might be 0110100110101101 and the next one might be 0011011010111000.  And this is done 44,100 times PER SECOND.

On top of that, the number of bits are doubled for a stereo recording.  So for the same of argument, your typical red book standard CD has a sampling rate of 44,100hZ and a 16-bitrate (sometimes expressed as 1411 kbps).

So yeah, that's a pretty good sampling rate and that's why CDs approximate an analog sample so well (except for that nasty HARMONICS problem, which I spoke of last time).

MP3 Conversion/Compression

I still think one great technological advance with huge impact was the development of the mp3 format to convert digital music.  Up until that time, digital audio files were in the 9MB range for a typical song.  With the old dial-up internet connections, it would take AGES to transfer one of these files.

Along comes a format whose files are 1/10th (on average) the size of an audio file.  A 9 MB file suddenly became a much smaller file and with advances in internet transfer, these songs could be moved in almost the blink of an eye.

Ahh, but how do preserve the digital signal when you have a file that is roughly a tenth of the size?  Well, you cheat a little.  Again, someone has decreed that since you don't hear much above 14000hz, then we can just LOP OFF the tops of those signals above 14,000hz and kind of approximate later, when we convert those mp3 files back to analog.

And hey...it works pretty good.  The standard bitrate for mp3 files appears to be 128 kbps (wow...did you just notice the difference between that and the CD bitrate, which was 1411 kbps??).  Now, if you're pumping that music through an ipod, through $14 ear buds, you ain't really gonna hear a huge difference.

But put that ipod on a nice Harman/Kardon dock on a nice amp and baby, you WILL hear the difference.

Of course, you could convert to mp3 of 196 kpbs (my personal favourite) or even 320 kpbs (best), but again, keep in mind that the bigger the bitrate, the bigger the file.

You might have to sacrifice some space on your iPod, but boy...your shit will sound great.

Anyway, that's my Sonic Physics lesson for today.

Here's MEESTER MUSIC's rules for digital audio file storage and playback:

I rip all my commercial (not downloaded) CDs to FLAC (free lossless audio codec).
big files (although not as big as wav files).  But no loss, no compression.  It copies EXACTLY what is coded onto the CD.

That becomes my storage "MASTER" if you will.  From there, I can convert those FLAC files to whatever I want:  back to wav to burn a disc, to mp3 (196 kbps) so I can play on my iPod.

That's it.  Enough of the techno stuff.  I start reviewing some of my latest acquisitions plus my favourite all-time LPs.


No comments: