Exporting 96khz file as 192khz

Hi,

I do most of my Audacity recordings at 24 bit, 192khz. Recently I needed to convert one of these files to 96khz because it was too big for the processing software I was using. Once I was done with the processing, I imported the file back into Audacity and continued working on it as 24 bit, 96khz. My question is whether there’s any benefit or harm to doing my final export to WAV as 24 bit, 192khz? It sounds fine to my ear, but I’m curious if doing this–resampling back from 96khz to 192khz–would introduce any artifacts or cause other subtle changes.

Thanks

Matt

Unless you are a very young woman, you are not going to hear any errors up in this realm – and sometimes even then. What you gain by running up that far is post production purity not absolute sound quality, and you’ve already kissed that off by downconverting. What you’ll get by upconverting is the conversion errors plus very, very accurate 96K sound errors.

I would stay there.

Koz

There should be no audible difference between exporting as 192, 96 or 48 kHz.
The sound may play a little better at 96 or 48 kHz as the DAC is not being stressed and has longer to accurately convert from digital to analogue.
Below 48 kHz you may notice a slight dulling of the sound. Above 48 kHz, any additional brightness or clarity is the result of conversion errors (distortion).
Converting between 193, 96, 48 kHz in Audacity is very high quality.

It is generally better to work in 32 bit float format as processing is more accurate. 32 bit float format is generally a bit more efficient on modern computers than integer formats and although the difference between 32 bit float and 24 bit integer is usually insignificant, the difference is cumulative so there may be a tiny audibly benefit in 32 bit float if you do a lot of processing or use multiple tracks. Converting from 24 bit to 32 bit float format is lossless (perfect).

Ok, got it. So once I’ve down converted to 96khz, stick with that the rest of the way.

Yes.
I’d also suggest testing recording at sample rates lower than 192 kHz. If possible, get someone to help you do a double blind A/B test. Higher numbers are not always better - it is quite common for A/D D/A converters to work better (more accurately) at more modest sample rates than at very high rates. Of course there is also the advantage that more modest rates use a lot less disk space, process faster and put less demand on disk I/O.

Thanks, I may do the double blind when I have time. Or perhaps I’ll do some digging around to see if anyone has already done the work for me: I have an M-Audio Audiophile 192 sound card, and perhaps someone’s compared it at different sampling rates. Anyway, thanks again Koz and Steve. I appreciate the fast replies.

Matt

My question is whether there’s any benefit or harm to doing my final export to WAV as 24 bit, 192khz? It sounds fine to my ear, but I’m curious if doing this–resampling back from 96khz to 192khz–would introduce any artifacts or cause other subtle changes.

There are teeny-tiny “mathematical” changes/errors, but you won’t hear any difference. If you go from 96 to 192 and back to 96, and the bytes will not be exactly the same as the original 96kHz bytes.

Going from 96 to 192 increases the amount of “data”, but it doesn’t increase the useful information… You can copy a VHS tape or DVD to Blu-Ray, and you get “higher resolution” but the picture won’t automatically improve.

Thanks, I may do the double blind when I have time. Or perhaps I’ll do some digging around to see if anyone has already done the work for me:

Yes, it’s been done before… As far as I know, nobody has ever “passed” a proper blind ABX test between CD quality 44.1kHz/16-bit and anything “better”.

There are “golden ear” audiophiles and “audio professionals” who claim to hear a difference, but the differences seem to disappear in blind testing… There’s always a “flaw” in the test, or they claim that blind testing is invalid for audio testing. :wink: The audio pros aren’t quite as goofy as the audiphiles (most pros don’t claim that expensive cables improve the sound), but there is a fair amount of nonsense (and lots of non-blind testing) in the pro audio community too.

In fact with most program material, most people can’t hear a difference between a high-resolution original and a high-bitrate MP3. (BTW - It turns-out that the equipment isn’t that important… If you can hear an MP3 artifact on a high-end system, you can probably hear it on an “average” system.)

24/96 seems to be the “studio standard”, and there may some advantages to recording, processing, and mixing at higher resolutions… Or, maybe not… But if your equipment supports it, there is no harm in using the higher resolution.

[u]HydrogenAudio.org[/u] has information on blind ABX testing and links to software that makes it easy to do yourself. But it can take some time since you have to do multiple trials in order to get statistically valid results. Of course if the difference is obvious, it shouldn’t take long to listen to 10 short samples and get 10 out of 10 “right”.

there is no harm in using the higher resolution.

Other than space. Computers are always a tradeoff between the room you have and the quality of the work.

For one popular example, when you’re running on the beach, you’ll never hear reduced sound quality over the seagulls and surf and you have a very reduced drive space and high need for different music, so highly compressed MP3 or AAC is indicated.

Koz