Home Computer Audio Asylum

Music servers and other computer based digital audio technologies.

Re: Why 48 kHz sample rate if destined for CD?

216.66.113.3

Thanks for the reply.

With respect to recording at 24 bits, the goodness of that choice was never in question. The added resolution is always a good thing and quantizing to a lower bit depth is (usually) fairly straight forward and less problematic than the downsampling problem.

As for the frequency extension of 48kHz vs 44.1kHz, that is also an unassailable point...if you have a device that plays at 48 kHz. If you want to record at 48 kHz because, in the future, you may use that format through your squeezebox, I think that that's a great reason.

My issue really does have to do with your points (3) and (4). Artifact-free downsampling from 48 to 44.1 kHz is not a trivial thing to do. In fact, with non-infinite bit depth it is probably not even possible. Some sort of distortion of the signal will always occur because of the finite-resolution math. With software that does the proper convolutions with a suitably high floating-point bit depth, though, you can get really really good results.

But, who knows what algorithms are actually being used in your boxed software? I know that CD Architect is a high-quality program, but how good is it with regard to this particular little nook of its operation? There are so many easier and obvious (and not so good sounding) ways of doing this downsampling. These other methods (linear interpolation, higher-order interpolation, interpolation followed by low-pass filtering, etc) are not just a little easier to implement, they are tremendously easier. And, if there was some programmer who's behind schedule and wants to get his manager off his back in a hurry, this would be a great corner to cut instead of taking the time to learn all the math and code up the that math into the software necessary to implement the "right" solution.

So, this is a long winded way of saying that, yes, you have good software, but you don't know if it's good in this feature. It may not be. In fact, for any boxed consumer/pro-sumer software, I'd argue that odds are against it.

So, that moves onto your fourth point regarding the professional's use of 48 kHz (again, 24 bit is a no-brainer. I've got no problem there). I believe that the history of "professional" use of 48 kHz has three factors. Again, this is my understanding of the history and it could be wrong...

First, it is my understanding that in the mid 1970's there were already a few existing digital audio formats...though used for non-music use. One influential standard was used in the telephone system and it operated at 8 kHz. As new digial audio formats were invented in the late 70's, their sample rates were often defined by the phyiscal medium on which they were recorded (often analog video tape, the source of the CD standard). But, if there was any choice in the mater, many of the new formats chose to be integer multiples of the earlier formats (such as the 8 kHz telephone format) because it makes cross-conversion easier and cleaner (which is my point exactly). Since 48 kHz is the first multiple of 8 kHz above 40 kHz (thereby allowing full 20-20000 Hz audio bandwidth), you get one reason for choosing 48 kHz.

The second factor for the 48 kHz "standard" flows from the popularity of DAT tape in smaller-to-midsize recording studios. I believe that DAT is generally the first widespread use of 48 kHz. Note that DAT (early 80s) came after CD (late 70s). The corporations that created the DAT standard wanted to make sure that DAT couldn't easily be used to make bit-perfect copies of their money-making CDs. They wanted to minimize piracy. Since 48kHz is hard to cross-convert with 44.1 kHz (again, this is my point), bit perfect copies would be hard to produce. Therefore, the choice of 48 kHz suited them well.

Finally, the third factor is that once some sort of "professional" standard was perceived (in this case, 48 kHz because of DAT), inertia took hold and, for the typical studio owner buying his first digital equipment, anything other than 48 kHz didn't sound like "professional" equipment. Then, once the studio owner finished mixing to DAT (or some other 48 kHz format), he shipped his tape off to the CD-mastering lab. There, the specialists took care of down-sampling his 48 kHz to 44.1kHz for CD. Since the CD-mastering people were specialists, they had the cash and the knowledge to get the properly designed hardware/software systems for doing this conversion well (Or, they were cheap and got the less expensive systems that employed simpler but crappier sounding algorithms).

In this day and age, though, people like you and I have to do our own Cd-mastering. Since you clearly like good sound (as evidenced by the excellent care with which you digitize your LPs), I would propose that maybe 48 kHz is not the best choice for you if you are going transfer to CD. I would propose that you try digitizing something in both 44.1 and in 48. Then, you can perform an expierment and decide if for you, with your particular soundcard, with your particular software, and with your particular sound system, if you prefer the sound of 48->44.1 or the sound of just 44.1. I'd be very interested to hear what you find. You might like the 48 better, you may not. I can't honestly say that I've ever heard a difference when testing with music. Typically, my kind of music and my stereo system (or my ears) just don't allow for that kind of refined analysis.

But I do find it fun to talk about. And, I've had to write software for non-integer downsampling before...so I like to talk about it A LOT. Thanks for giving me an opportunity to discuss its finer points. It's a fun hobby!

Chip



This post is made possible by the generous support of people like you and our sponsors:
  Sonic Craft  


Follow Ups Full Thread
Follow Ups


You can not post to an archived thread.