Sample Rate Conversions - hosts compared!

[nuisance sonore]

This makes sense. I do the same thing with photography. I upscale 8bit per channel images to 16bit and when moving pixels, double the image size. Then do significant editing, the downsample/scale.

Sorry but your analogy is wrong. Going from 8 to 16 bit in an image would be similar to going from 16 to 24 bit in audio as your adding more numbers to describe each pixel, therefore more precision and can be beneficial, but not adding pixels artificialy.

Going from 44.1KHz sample rate to 88.2KHz by upsampling would be like taking an image that's 400x400 pixel and resampling it to 800x800 pixels before working on it, only to resample it back down after. In photo work, it's very destructive to do so and nobody works like that.

bmanic, I would trust my ears to try, say, a compressor and see if I like it. But to apply such a change to methodology I wouldn't trust my ears, I need to know the "why" before I apply complete change to my work method. I'm not dismissing the potential benefit of working like this, but I'm not going to unless I understand the logic behind such a method.

[jmh]

Going from 8 to 16 bit in an image would be similar to going from 16 to 24 bit in audio as your adding more numbers to describe each pixel, therefore more precision and can be beneficial, but not adding pixels artificialy. In photo work, it's very destructive to do so and nobody works like that.

Actually some of us do that when working with graphics. Especially if one is working in the oh so wonderful world of web graphics where clients send you thumbnail sized images that you're supposed to use in the layout. Creative trickery in such cases is one of those things I use in my mind to separate the fiddlers from real graphicians.

Elsewhere in the graphics world, it's quite common (although by no means universal, as graphics world is at least as full of selfclaimed designers as the audio world is full of "producers") to work at a higher resolution and downsample the results. Of course, this is different from starting with a lores original and upsampling it first.

JMH

[OMU]

to apply such a change to methodology I wouldn't trust my ears, I need to know the "why" before I apply complete change to my work method. I'm not dismissing the potential benefit of working like this, but I'm not going to unless I understand the logic behind such a method.

Mastering processors are designed to work best at higher sample rates.
Some of them even resample by themselves the file before processing it (UAD, PSP etc).

[Kingston]

I would be interested to see a similar scientific analysis and comparisson of bit-rate conversions/dithering, which most of us probably do more regularly.

There has been a really good graphic comparison available for a long time. I think it was done in connection with the research of the megabit max dither that now resides in the ozone processor (it's the best dither).

I'll dig up the article and post it here.

[shamann]

Are there bad dithering schemes out there? Ones most obviously to be avoided?

[Kingston]

Not really. The tests quite clearly pointed the differences to be very minor. We're talking about shuffling 1bit here.

While the difference can be heard with the best converters, it isn't significant enough to be worried about (especially if you already have ozone).

The obvious thing to worry about is to not accidentally use cascaded 16bit dithers, or change the volume after dithering.

With 32/24, all this becomes irrelevant.

I can't find the link I mentioned. If anyone knows the one I'm talking about and still has it bookmarked, don't hesitate to post here.

[Kingston]

Found it!

http://audio.rightmark.org/lukin/dither/dither.htm

The graphs don't load here, but there's plenty of information regardless of it.

Oh and by the way, this test is old, but there hasn't been any improvements in the field of dithering since megabit max.

[headquest]

Thanks for that Kingston. I'll check it out later. I do all my dithering in Audition, which I assume will be OK given the general high regard it is held in, but I have actually been considering Ozone as well... and noticed that the price seems to have come down quite a bit recently...

[Teksonik]

All this so someone can play the end results squashed down to mp3 thru earbuds on his iPod.

[Kingston]

All this so someone can play the end results squashed down to mp3 thru earbuds on his iPod.

[deaf dunderkwac]

All this so someone can play the end results squashed down to mp3 thru earbuds on his iPod.

But this is hardly the same as dubbing down to 8 track or cassette.

[SJ_Digriz]

All this so someone can play the end results squashed down to mp3 thru earbuds on his iPod.

But this is hardly the same as dubbing down to 8 track or cassette.

yep, it's far worse going to mp3

[Hink]

All this so someone can play the end results squashed down to mp3 thru earbuds on his iPod.

But this is hardly the same as dubbing down to 8 track or cassette.

yep, it's far worse going to mp3

not really, mp3s don't wear out

[deaf dunderkwac]

and there was nothing more excruciating then having your carefully prepared tapes sound mangled by a cassette with Dolby B 'on' (nevermind the wow and flutter, noise and generally high levels of distortion)

at least a high bit-rate mp3 sounds remarkably simular to what you started with!

[jasonsantiago]

A guitar recorded at 44.1 sounds better in relation to 88.2 than a vst synth. Software synths sound clearly better at the higher sample rate because you lose the aliasing artifacts.

you can alias recording anything doing A --> D conversion, even guitars...

also, these tests mean just about nothing to me. If you can't hear it, who cares? Also, in the case of some analog processing, some non-linear harmonic distortion is considered 'pleasing' to the ears, so who knows what the tests mean to our ears?