44.1 kHz is the frequency used by CDs, so yes, audio cards can handle it. But 48 kHz is much better for video because if you divide 48000 by 24, 25, 30, 50, 60, you get an integer. But if you divide 44100 by 24 you do not get an integer (you do with those other frequencies).

So, with 48 kHz you get an integer number of samples per video frame. With 44.1 kHz you may or may not get an integer number of samples per video frame.

At any rate, resampling 48 kHz down to 44.1 kHz makes no sense. A video is not a CD. Most video players are optimized for 48 kHz and even 96 kHz, but not 44.1 kHz.

Now, if he wants to sell a CD with the music of his video, he will have to resample the music to 44.1 kHz for the CD. But he should still keep it at 48 kHz with the video.

I hope that when he does resample 48KHz to 44.1KHz using Vegas he knows to set Resample Quality in his project to Best. The Default "Good" is not very good.

Bob.

I'm not saying it was right, but I know Youtube recommended 44.1Hz audio when uploading videos to them, for the longest time, but it doesn't say that anymore. Now they'll take anything.

You are correct, but 48K worked just fine on Youtube then, too. If I've ever rendered audio-for-video at 44.1, it was by mistake.

FWIW, Youtube has been changing its APIs and renderers again, and they are reportedly now using x264 for HD, according to one of the developers at Handbrake. Should be good news if they're using it sensibly, and not clobbering the MBR like they did with their older proprietary encoder.

Musicvid: He also works as a DJ and apparently believes that 44.1 is more pleasing to the ear, but the videos in question do not contain music anyway. He also claimed that both Vegas and Premiere have 44.1 audio as the default in many of their presets for video, so it must be correct. I felt that leaving it at 48 was better, but have to question why Sony and Adobe both do that.

Red Prince: If you don't mind my asking, what is the advantage of having the audio match up with even integers - better sync with the video frames?

A default setting is not better, its just a default. 48K for consumers began with DV video fifteen years ago and has been more or less the norm ever since.
For pristine results, 44.1 should encode at 44.1, and 48 should encode at 48.
If there is a mix of the two on the timeline (such as an added mp3 for background), I render at 48K to preserve the higher quality audio. The mathematical difference is of course 8% of samples that would be lost to downsampling. Dithering applied to upsampled audio is rarely a problem, but one should be aware.

That all being said, if it's just speech in your program, it's not going to make a whit of difference.

Yes, easier sync, though it would be arguable with NTSC, which is not exactly 30 fps but 30/1.001 fps.

Also, his belief that 44.1 kHz is somehow more pleasing to the ear than 48 kHz makes about as much sense as saying that 480p is more pleasing to the eye than 720p.

In reality, with decent sound equipment 44.1 kHz and 48 kHz will at best sound the same to the ear because both are above the Nyquist frequency. And with a less decent sound equipment, if there is any difference between the two, the 48 kHz is better.

I have seen sound people arguing over whether they sound equal or whether 48 kHz sounds better than 44.1 kHz. But I have never seen anyone arguing that 44.1 kHz was better of the two.

Because of the Nyquist frequency mentioned above, the highest frequency that the 44.1 kHz sample rate can handle without aliasing is 22,050 Hz, while the highest frequency the 48 kHz can handle without aliasing is 24,000 Hz. Both are higher than the highest frequency humans can hear (20,000 Hz when we are young, much less as we age), which is why some sound people argue they sound exactly the same. But just because it is theoretically possible to reconstruct the same analog sound wave from either sampling rate, it does not mean that all sound equipment is up to what is theoretically possible. Which is why others insist they can hear the difference between the two and the 48 kHz sounds better to them.

According to legend, the size and sampling rate of the CD were chosen to fit Beethoven’s Ninth Symphony on one side of the disc. Had they chosen 48 kHz, it would not fit. Had they chosen 24 bits per sample, it would not fit. So, to fit it all on a 12-cm disc, they chose 44.1 kHz at 16 bits per sample. So much for it being chosen for pleasing sound. :)

Personally, whenever I can, I record at 96 kHz/24 bits. Then when I mix multiple sound tracks, I get lesser round-off error before the final conversion to 48 kHz than I would have if I recorded at 48 kHz from the start.

Then the dude is an idiot. Do not take anything he says seriously.

FWIW I am primarily an audio user, and record stuff for CD at 44K1, but that is only because it will end up at 44k1 and resampling should be avoided if possible, if not for denegration of quality, just because 'why do any extra unnecessary step ?'

geoff

48k has always been the standard for audio with digital video. Anything else is considered non-standard by the industry.

Sony did have cameras that recorded at 32k. Don't know why but Sony always danced to its own drummer.

Stick with 48 unless you
1) need to make a file smaller and audio quality can be sacrificed
2) want other programs to resample back to 48k
3) just want to be different.

Sony Vegas started out as an audio editor, so for some reason they have always left 44.1 as their default.

Dave T2

44.1 was chosen for the new CD standard not necessarily because anything higher wouldn't fit, but because that was the sampling rate then used by the only other common digital recording format--PCM on videocassette.

Great Wikipedia article here:

http://en.wikipedia.org/wiki/44,100_Hz

I can sometimes hear Q-noise on a CD that I can't detect at 48K.
I don't know that I could identify each in a blind test, but some CDs are annoying because I'm particularly sensitive. It's the same reason I stopped doing MIDI projects.
And those over-compressed TV ads! -- well, that's a different rant.

Quantisation noise has absolutely nothing to do with sample rate. If you can hear it on some CDs, then there is something badly wrong in the recording or mastering of those CDs.

If you really want to hear some quantisation noise, listen to Ry Cooder's Bop Till You Drop , on CD or vinyl ;-)

geoff

The more slices that are sampled, the fewer rounding errors, and more accurate regression. it's just math.
But perhaps at some specific frequencies this does not always hold true because of spurious harmonics.
I'd be interested in seeing the source of your information. Always willing to learn.