sample rate conversion question

[federal donut]

Should you convert down to 44.1 from 96 after dither from 24bit to 16 or before?

[stillafool]

Before. You want to add dither last.

[Piotr]

If possible in the future, better to record at 88.2k and then sample convert to 44.1k...

[JamesHE]

I don't buy into this whole division by 2 thing. Possibly somwhere, yeah maybe some decimal places don't get chopped off. I really don't have a clue as to how the resampling works. But If you have a resampling algorhythem that's worth a shit I really don't think it's going to care if it's dividing by 2 or some other real number. I bought into that far a while, but I haven't in my own listening come to the same conclusion.

[kcrusher]

The division by 2 thing is complete, total, utter BS. Please stop propagating this myth, as it's completely untrue.

There's alot more to sample rate conversion than just 'throwing away every other sample'. Typical algorithms recreate the waveform from the original digital recording and then resample it at the new sample rate.

[drliebs]

I just resonded to another thread with a similar question.

I don't know why my audio sounded different but it did. It came form a Tascam HD24 recorded at 48k SDII files. I converted the files in Pro Tools to 44.1 wav files. The high end was definitely effected. I don't know why, but I could hear it. I wish I knew if it was something I did wrong.

On one side, the argument is record at high sample rate so any processing will sound better , even if it gets converted to a lower rate for CD. On the other side, there is the oppinnion that if it is going to end up at 44.1 16 bit for CD then start that way and leave out any conversion process.

And then there is 192k...some people say it is overkill, others swear by it. The debate goes on.

[cassembler]

The division by 2 thing is complete, total, utter BS. Please stop propagating this myth, as it's completely untrue.

I _promise_ I'm not trying to be a dick, but read Ken Pohlman's Principals of Digital Audio and tell me that statement again. Or better yet: Record a full set of drums at 88.2 and 96, downsample both to 44.1 and see what sounds better. (btw, the always dither after sample rate conversion advice is right on, but I think it's only necessary when also doing a bet depth change)

BTW, 44.1 is by no means a wonderful sample rate, it just so happend the technology in the CD era wasn't designed to be real future friendly.

$.02

[cassembler]

To add to the above, I'd personally take a higher bit depth over a higher sample rate any day of the week. This ONLY applies, again, if the material will only be released on CD.

If you have the budget (storage and processing-wise), working at a higher sample rate is a good idea for the future of audio, when CD's go the way of the 8 track.

PS: you're right in saying that 88.2 to 44.1 isn't as easy as just removing every other sample, there's a re-interpolation process as well, it's just that it _theoretically_ is a simpler math than going from 48 to 44.1, and the re-averaging process is pretty much perfected. I dunno.

$.02

[cassembler]

(This was posted in the other thread)

All of this depends directly with the quality of the resampling algorithm; I suspect 90+% of the people on this board (myself included) could not detect the difference between downsampling 96 and 88.2 to 44.1 ASSUMING THE CONVERSION WAS OF VERY HIGH QUALITY.

Hmmm.... perhaps I must qualify my stance on the subject to include "it's better to track at 44.1 the whole way, unless you're positive that your sownsampling is of very high quality."

[kcrusher]

The division by 2 thing is complete, total, utter BS. Please stop propagating this myth, as it's completely untrue.

I _promise_ I'm not trying to be a dick, but read Ken Pohlman's Principals of Digital Audio and tell me that statement again.

I stand by my statement in full. Downsampling from 88.2 to 44.1 does NOT simply involve throwing away every other sample. I admit I have not read Ken Pohlman's book, but if this is what he is saying, then I would call him to task on his actual knowledge of digital audio.

Or better yet: Record a full set of drums at 88.2 and 96, downsample both to 44.1 and see what sounds better. (btw, the always dither after sample rate conversion advice is right on, but I think it's only necessary when also doing a bet depth change)

Ok, even more myths going on here. You NEVER dither when doing a sample rate conversion, only when doing a bit depth change.

Your subjective experience with downsampling is just that - subjective. I've done plenty of recording at higher sample rates and downsampling and I can honestly say that, if the downsampling algorithm is good, there should be very negligible difference between them. Not to mention that there are many other factors that may have played a part in the subjective experience - unless you recorded the drummer simultaneously with the same mics, you still would have had to use different convertors to capture the 88.2 and 96k audio files. Unless they were high end convertors, there may have been a difference between the convertor quality (even the same mfr/model can have enough differences to be noticeable). The levels during recording would have had to have been matched to within 1db to even be considered remotely equal and even that is a generous figure.

I would rethink your stance on this before continuing.

[Piotr]

Everyone likes to do things their own way...I know that things don't sound much different when you downsample from different rates. I just like to keep things neat and tidy!

[joel hamilton]

Regardless of the "math" involved, 88.2 sounds almost exactly the same as 96k to my ears, though 96 has a weirder overall "ring" to it in the way high end. I have tried everything at this point from 44.1, to 192k, and I really like 88.2 because it sounds about the same as any of the "HD" formats, and takes up less drive space. I am much more concerned with the 80 bazillion OTHER variables in the recording process.

Not one record will ever come out, and have a reviewer or consumer say "man this could have been a hit, but they didnt go 88.2 instead of 96k when they tracked it." Never. I still listen to some really good songs I tracked at 16/44.1 on a blackface adat...

[Electricide]

is anybody concerned yet with the Next format? we're all still slaving to the 44.1 CD god, but really, with SACD, or DVD-A near, how long will you need a 44.1 version? It is still the cheapest way to do a one off for a friend, but if your indie band is posting on the internet anyway where it will be killed in compression.....why not do the highest resolution possible? So it will downsample better to a CD? Why not think about not having to upsample in a few years.

The sooner we stop using the CD format as a major factor in choosing a sample rate, the sooner that format can pass into history. It is finally cheap and accessible, but that's because the technology has moved on already.

[drliebs]

I have had this debate with some folks who agree that it all ends up at 44.1 in the end and CD's are going to be around for a long long time coming.

It has been compared to digital photography. If your final digital picture will have a resolution of 640 pixels by 800 pixels, what is the point of doing photo re-touching and stuff on a larger version of the file when it is just going to be resized in the end. It makes the processor work harder, it takes more space, so you can't do as much at once, it takes longer. But, and this is a big but (I like em' and I cannot lie) all of the retouching and processing looks much better on a larger picture, and when it get's resized it will still look better than the technique of doing all of the processing on a smaller sized (lower resolution) file. Working in high res, there is more detail to work with so all of the filtering just wokrs better. It has more to chew on,

This may be going in a weird direction, it is early in the morning, but I think the same can be said of audio. The final product may be a lower bit rate or it may even end up as an mp3, but working in higher resolution will give you a better sounding product in the end. Everything you do to the audio after it has been digitized will sound better int the higher reolution.

Not one record will ever come out, and have a reviewer or consumer say "man this could have been a hit, but they didnt go 88.2 instead of 96k when they tracked it." Never. I still listen to some really good songs I tracked at 16/44.1 on a blackface adat...

This is so true, how many people think mp3's sound "just as good as the cd" This is what ties us down to CD's for a long time. Most people don't care enough to spend the money to buy a SACD or 24 bit CD player, and why would they when an mp3 is "just as good".

For personal stuff, demos and things like that I do at at home, I track at 44.1 In the studio I track at 96k. I think 192 is overkill but I am just a semi-pro engineer and I am not going to say I know more than someone else. Just my oppinnion.

[kcrusher]

SACD and DVD-A are going to be esoteric for the near future. You'll have to keep in mind that they won't become a 'standard' until a vast majority of the population have decks to play them on and one of them emerges as an actual 'standard'. Most people really don't care too much about sound quality - as evidenced by the popularity of MP3's. Until people see (hear?) a clear cut improvement, it's unlikely that the traditional CD is going to be usurped.

You'll see more 'high-res' releases as the days go by, but the tried and true will still be around for quite awhile - at least another couple years or so.

The reality is that and decent home rig is more than adequate for making a 'professional' release, as long as the person behind it knows what they're doing. All this extra fancy schmancy gear is really there to please ourselves and those around us who we know have good ears and can appreciate great sound quailty. The masses couldn't give a rats ass what you recorded it on - they just want to hear a good tune.