How Is Mu­sic Stored In the Dig­i­tal For­mat and What Do Higher Res­o­lu­tions Mean?

NOVO - - 16BITS VS 24BITS - Mal­colm J. Gomes

While it may be true that vinyl is mak­ing a come­back, in to­day’s world of mu­sic, dig­i­tal en­joys to­tal dom­i­nance. Dig­i­tal mu­sic made its de­but at the con­sumer level in the early 1980s with the in­tro­duc­tion of the com­pact disc sys­tem and the claim of ‘per­fect sound for­ever’. Given that the rst dig­i­tal play­back sys­tems de­liv­ered sound that was hard and edgy, that claim was ob­vi­ously over the top hype. How­ever, thanks to ad­vances in dig­i­tal tech­nol­ogy and es­pe­cially the big strides made in dig­i­tal to ana­log con­ver­sion, es­pe­cially over the past decade, we can now en­joy dig­i­tally re­pro­duced sound that ap­proaches the smooth, silky prop­er­ties of vinyl re­pro­duc­tion.

For the long­est time, dig­i­tal mu­sic was as­so­ci­ated with the 16-bit, 44.1 kHz con gu­ra­tion of the com­pact disc. Over the past few years we are see­ing more mu­sic be­ing of­fered in higher res­o­lu­tions like 24-bit and up to 192 kHz. So, does higher res­o­lu­tion au­to­mat­i­cally re­sult in bet­ter qual­ity sound re­pro­duc­tion? The short an­swer is that it de­pends on many fac­tors.

To be­gin with, let’s un­der­stand what these ‘bits’ are all about. Many peo­ple re­fer to these as the bit rate, when a more ac­cu­rate de­scrip­tion would the bit depth. Bits are sim­ply bi­nary in­for­ma­tion (ze­ros and ones) that forms the data, which stores the mu­sic. The bit depth tells you the num­ber of bits that are em­ployed to store the au­dio sig­nal.

The process of stor­ing mu­sic in a dig­i­tal for­mat in­volves slic­ing the au­dio sig­nal and stor­ing each slice as a bi­nary code. When it is done in the 16-bit for­mat, there are 65,536 lev­els. With each ad­di­tional bit depth, this num­ber dou­bles, so when you get to the 24-bit for­mat, you now have 16,777,216 lev­els.

If bits re­late to lev­els, then the sam­ple rate re­lates to time. In other words, the sam­ple rate tells you how many times the au­dio sig­nal is mea­sured or sam­pled per sec­ond. In the case of the com­pact disc, the sam­ple rate is 44.1 kHz, which means that each sec­ond of the mu­sic is di­vided into 44,100 slices. Higher sam­pling rates like 96 kHz would there­fore mean that each sec­ond of mu­sic is di­vided into 96,000 slices. In the case of com­pact discs, the 16-bit depth and 44.1 kHz trans­lates to each sec­ond of the mu­sic be­ing di­vided into 44,100 slices and each slice con­tain­ing 65,536 lev­els.

As you know, all this bi­nary infor- ma­tion has to be stored ei­ther on a disc or on a drive. It was not long ago when most hard drives sold at the con­sumer level were in hun­dreds of gi­ga­bytes. Now we have af­ford­able hard drives that of­fer stor­age of one or more ter­abytes (one ter­abyte is a thou­sand gi­ga­bytes). So how do you re­late bit depth and sam­pling rates to the space on a hard drive?

For that you need to con­vert the bit depth and sam­pling rate into Mbit/ sec. The math for this con­ver­sion is a bit com­pli­cated so in­stead let me give you a few ex­am­ples that will give you an idea. If the bit depth is 16-bits and the sam­ple rate is 44.1 kHz, the Mbit/sec is 1.35 and for one minute of a stereo sig­nal you will need 10.1 megabytes of space on a hard drive. Since most songs are around 3 min­utes long, each song, on an av­er­age will need 30.3 megabytes of space on a hard drive.

Lets com­pare this to an MP3 le with a sam­ple rate of 128. Here the bit rate is just 0.13 Mbit/sec for one minute of a stereo sig­nal. You will there­fore need just 0.94 megabytes of space for each minute of mu­sic and an av­er­age song will take up just 2.82 megabytes of space.

In con­trast, a 24-bit depth and 96 kHz sam­pling rate makes the bit rate shoot up quite ex­po­nen­tially

to 4.39 Mbit/sec which in turn re­quires 33 megabytes for each minute of mu­sic and an av­er­age of 99 megabytes per song.

It is im­por­tant to note that when you record and rere­cord mu­sic on a hard drive it may not be done in a lin­ear method. The hard drive may al­ready con­tain songs that have been recorded pre­vi­ously and these may not be recorded neatly one af­ter the other on the drive. This will re­sult in gaps be­tween the songs that may be left blank when you record more songs on the drive. A lot of empty space could also be left on the drive based on the disk’s sec­tor size.

Given that 24/96 res­o­lu­tion takes up so much more of your hard drive space for each song when com­pared to MP3 and 16/44.1 res­o­lu­tion, is it worth con­vert­ing and stor­ing your mu­sic in 24/96 or higher res­o­lu­tion?

From the tech­ni­cal point of view the ad­van­tage is that your au­dio in 24/96 will de­liver res­o­lu­tion that is 250 times greater than the 16/44.1 that com­pact discs de­liver. The dis­ad­van­tage is that a 24/96 le will take three and a quar­ter times the space on your hard drive as com­pared to 16/44.1.

Right off the bat, let’s dis­card the no­tion that 250 times the res­o­lu­tion means 250 times bet­ter qual­ity sound. A lot will de­pend on how good your hear­ing is and the qual­ity of the other com­po­nents of your au­dio sys­tem, es­pe­cially the dig­i­tal to ana­log con­verter (DAC) that you em­ploy. If your hear­ing is not up to scratch, and if your other com­po­nents are just about av­er­age or be­low av­er­age, chances are that you may not even notice a dif­fer­ence be­tween 16/44.1 and 24/96. Based on hu­man hear­ing, you are a lot more likely to notice the dif­fer­ence be­tween an MP3 record­ing and its 16/44.1 coun­ter­part.

Which brings me to why 16/44.1 was se­lected for the com­pact disc for­mat. This res­o­lu­tion was cho­sen based on the Nyquest The­ory. This the­ory states that the ac­tual up­per thresh­old of dig­i­tal au­dio will top out at half the sam­pling rate used. Since hu­man hear­ing usu­ally tops out at ap­prox­i­mately 20 kHz, se­lect­ing 44.1 as the sam­pling rate would, based on this the­ory, top out at around 22 kHz which would com­fort­ably ac­com­mo­date all the sound fre­quen­cies that are au­di­ble to the hu­man ear.

If this is the case, why even bother with higher res­o­lu­tion for­mats like 24/96 and higher? The an­swer to that can be found in the vinyl record for­mat, which has been mea­sured to con­tain fre­quen­cies as high as 50 kHz. Some vinyl a ciona­dos claim that the su­pe­ri­or­ity of vinyl sound re­pro­duc­tion over dig­i­tal (that they hear) is be­cause the much higher (in­audi­ble) fre­quen­cies that the vinyl sys­tem re­pro­duces have a pos­i­tive ef­fect on lower (au­di­ble) fre­quen­cies of the mu­sic which, they claim, ac­counts for the su­pe­ri­or­ity of vinyl re­pro­duc­tion.

If this is in­deed true, then us­ing the Nyquest The­ory, the same higher (in­audi­ble) fre­quen­cies can be re­pro­duced us­ing the dig­i­tal do­main by rais­ing the sam­ple rate to 96 kHz, which would re­sult in the ca­pa­bil­ity to re­pro­duce sound sig­nals with fre­quen­cies as high as 48 kHz. It needs to be said that this claim has not been con­clu­sively proven and needs to be taken with a pinch of salt.

An­other ex­pla­na­tion as to why 24-bit is bet­ter than 16-bit is that the for­mer may not nec­es­sar­ily im­prove the sound qual­ity di­rectly, but that it gives the au­dio sig­nal more room to breathe in the nu­meric sphere of dig­i­tal au­dio. It is claimed that this makes it pos­si­ble to record mu­sic with greater dy­namic range where the softer pas­sages are qui­eter and bet­ter able to stay above the noise oor and the louder seg­ments of the mu­sic are re­pro­duced more cleanly and with­out clip­ping. If this is true, then the re­duced noise would make it pos­si­ble to record at lower lev­els so that you en­joy more head­room and in­stru­ments sound cleaner, vo­cals clearer and the higher fre­quen­cies ren­dered with more del­i­cate­ness. There is also a school of thought that a sam­pling rate of 88.2 sounds bet­ter than 96 be­cause it is an ex­act mul­ti­ple of 44.1. Again, there is no con­clu­sive proof that this is true.

Hav­ing said all that, we hu­mans are blessed with an in­cred­i­bly sen­si­tive sense of hear­ing and so it may not be a bad idea for you to use your ears to be the nal ar­biter in this mat­ter. If you can clearly and con­sis­tently hear the dif­fer­ence be­tween a 16/44.1 and a 24/96 le of the same mu­sic, then, by all means, go for the lat­ter. You would also be well ad­vised to ex­per­i­ment with preamp and DAC up­grades, be­cause some­times this might be a more cost ef­fec­tive route to bet­ter sound qual­ity than the on­go­ing ex­pense of grow­ing your 24/96 mu­sic li­brary.

As for even higher sam­pling rates and bit depths like 192 kHz and 32 bits, it would be worth not­ing that mu­sic les in these res­o­lu­tions are huge and even with a hard drive of a cou­ple of ter­abytes, you will run out of disk space rel­a­tively quickly even if you have a fairly mod­est dig­i­tal mu­sic li­brary. Be­sides, most au­dio­philes, and that in­cludes me, can­not con­sis­tently tell the dif­fer­ence be­tween a 24/96 and a 24/192 sam­ple rate le of the same mu­sic. It does seem like many 24/192 mu­sic les are more a one-up­man­ship mar­ket­ing ploy to charge more rather than some­thing that of­fers real value.

Mu­sic fans should also be warned when buy­ing mu­sic les that have a higher res­o­lu­tion than 16/44.1. There have been in­stances in the re­cent past where even some of the most re­spected brands have sold high-res­o­lu­tion mu­sic les that were lit­tle more than up-sam­pled 16/44.1 les. This is a to­tal rip-off as res­o­lu­tions higher than 16/44.1 are the real McCoy only if they are na­tive high res­o­lu­tion rather than an up-sam­pled ver­sion of a lower res­o­lu­tion for­mat.

Fi­nally, you are likely to come across claims of 32-bit oat point pro­cess­ing. So what on earth does that mean? Many con­sumers tend to con­fuse this with 32bit record­ings when in re­al­ity it refers to the fact that some ma­jor se­quencers and many of the bet­ter multi track recorders are ca­pa­ble of ren­der­ing au­dio tem­po­rar­ily in a 32-bit oat­ing point for­mat. The ra­tio­nale be­hind 32-bit point pro­cess­ing is that if prop­erly im­ple­mented, it adds ex­tra bits to the mu­sic le af­ter record­ing to al­low more head­room for au­dio math­e­mat­ics in the dig­i­tal do­main. Be­fore the le is out­put it goes through con­vert­ers to bring them back to the orig­i­nal 24-bits. You can think of the “ oat­ing point” as a scal­able dec­i­mal point in a cal­cu­la­tion. Since you have 32 rather than 24 reg­is­ters for cal­cu­la­tions, it is go­ing to ren­der a more ac­cu­rate re­sult. The other ad­van­tage of 32bit oat point pro­cess­ing is that it makes some cal­cu­la­tions pos­si­ble that would, in the­ory, be im­pos­si­ble with a 24-bit con gu­ra­tion.

Now that you have a bet­ter un­der­stand­ing of how mu­sic is stored in the dig­i­tal do­main, I en­cour­age you to ex­plore higher res­o­lu­tion mu­sic – only then you’ll be able to de­ter­mine if it pleases your ears.

Newspapers in English

Newspapers from Canada

© PressReader. All rights reserved.