Comparing Sample Rates

[DJ Insomaniac]

Hey, I'm looking for some pointers on how to compare the sound of a 24-96k file to a 16-44.1k.

I did a session where I recorded at 96k, then exported it as a 16-44.1, then imported it back in PT (hoping that the resampling up to 96k didn't change much). I'm really curious to hear the difference made by the higher speed, but I the two files seem to sound the same. Has anyone ever done this, and if so, what did you use as a sound source? It would seem that the change wouldn't be as drastic as going to 8-bit audio, so what should I be listening for?

This is a little experiment I've been wanting to do for a while, so any pointers would be very welcome.
Thanks!

[DJ Insomaniac]

bump.

Anyone?

[Slim Shady]

Sounds to me like you answered your own question...

[Lapick]

Greetz DJ Insomaniac,

There is a difference between 16/44.1 and 24/96 to be sure but depending on the source material and other factors like the quality of the playback system, dithering method, etc. it may be more subtle than you'd think. Having said that I'll add that I've done similar tests and have absolutely heard a difference.

Instead of starting with the same 24/96 session I'd suggest doing one session at 16/44.1 and another at 24/96 using the same source material. A familiar track from a favorite CD is good because it doesn't introduce performance variables that could be psychoacoustically interpreted as differences in quality. Just be sure to pick a well mixed and mastered CD and a track that has lots of detail like crisp vocals, cymbals, clear reverb tails and such.

As for what to listen for pay particular attention to imaging and depth or dimensionality, all of which are dependent on the integrity of sonic detail. If you include the original reference track and listen blindly (you may need a friend to help you with that) it might make the test a little more interesting.

Have fun and let us know what you conclude.


All the best,

Lapick

[DJ Insomaniac]

Cool, thanks.

Would an analog source be a good bet, like vinyl? Wouldn't a CD just be 16/44.1 sampled at 24/96k, or does it not really matter?

I feel so silly having to ask these questions, but i spent a while trying things and not really seeing a result.

[Jon_Atkinson]

There is a difference between 16/44.1 and 24/96 to be sure but depending on the source material and other factors like the quality of the playback system, dithering method, etc. it may be more subtle than you'd think. Having said that I'll add that I've done similar tests and have absolutely heard a difference.

Instead of starting with the same 24/96 session I'd suggest doing one session at 16/44.1 and another at 24/96 using the same source material. A familiar track from a favorite CD is good because it doesn't introduce performance variables that could be psychoacoustically interpreted as differences in quality. Just be sure to pick a well mixed and mastered CD and a track that has lots of detail like crisp vocals, cymbals, clear reverb tails and such.

As for what to listen for pay particular attention to imaging and depth or dimensionality, all of which are dependent on the integrity of sonic detail. If you include the original reference track and listen blindly (you may need a friend to help you with that) it might

Are you suggesting taking a 44.1kHz 16bit file, and playing it back in a PT session at 96kHz 24bit, then comparing it to the original file?

[Lapick]

Sure, a good vinyl recording played on a good table would be great. The goal is to start with the best source possible while maintaining consistency. I suggested a pre-recorded source for that reason but you certainly could use a live source like voice, guitar, etc. You just need to make the performances as identical as possible and that's easier said than done.

[Lapick]

Hi Jon,

No, I was suggesting that he record the same source through his converters into seperate sessions at different resolutions. I assume what he's really interested in is the difference between the two resolutions as recorded on his system in particular which would need to include his converters. Comparing them to the original can also be instructive as his converters and associated analog circuitry will certainly color things in some fashion.

Lapick

[Jon_Atkinson]

No, I was suggesting that he record the same source through his converters into seperate sessions at different resolutions.

Ok.... that makes sense.
In answer to the original poster's question regarding the audible difference between sample rates and bit depths, here's the story (very much simplified...!!!)

The difference between 16bit and 24bit is pretty large.
The dynamic range of a 16bit system is 96dB, the dynamic range of a 24bit system is 144db. In real life how this 'sounds' is that 16bit recording at lower levels can sound pretty thin, whereas 24bit can be pushed much less, and still sound great.

Recording at different sample rates is a different story, and is determined far more by the quality of the converters.

For instance, in theory a perfect AD converter recording at 44.1kHz, and then playing back through a perfect DA would create exactly the same sound as one recording at 96kHz.
Of course converters are not perfect, and neither are DAWs. However converter technology has come on in leaps and bounds in the last couple of years, to the extent that the best converters out there do sound exactly the same at 44.1 as at 96. Anti aliasing filters in better converters are more accurate, and can start at higher frequencies, thus reducing the 'harshness' often associated with older digital recordings.
There are also issues with how well plugins behave at different sample rates, which can also confuse the argument.

In essence though, the movement towards higher sample rates was fuelled by poor performing converters (which can sound better at higher rates), but due to improvements therein the upward pressure for this has abated (thank goodness!!).

Remember that the human ear can only hear up to 20kHz (in reality probably only 18kHz), so in actuality 44.1kHz sampling rate should be ample according to Nyquist.
In reality 48kHz can be a better bet, as the anti aliasing filters will cut in at a higher sample rate (perhaps 22kHz instead of a possibly audible 18kHz), but the difference between 48 and 96 is in theory non-existent. If you do hear a difference performing the test suggested above, then it's time to buy some better converters, and record at 44.1 or 48.

I appreciate this will probably start a 'flame-war' with posters claiming there are massive audible differences between sample rates. That's fine, if you do believe you can hear it.
The laws of physics say that there is no audible difference.
The laws of reality regarding DAW and plugin behaviour say that there might be a difference, but it's probably not worth losing sleep over.

[Lapick]

Flame-wars, here... Never!