Axe as audio interface... Flexible sample rate?

[Wolfenstein98k]

I just have to comment on this. Whether, you are sampling at 48 kHz or 96 kHz has nothing to do with, if you record pitches higher than 20 kHz. It has to do with the amount of harmonic content you pick up and the ability to regenerate the incoming waveform.

If you record a 20 kHz tone at 48 kHz, you get two sample per period (one full evolution of a sine wave). So when the digital signal has to be converted back to the analog domain, it can only assume that the incoming signal was a perfect sine wave, which may or may not be the case. At 96 kHz, the A/D converter picks up four samples (almost five) per period. This gives the D/A converter much more information to work with when regenerating the analog signal.

All that said, I still record at 48 kHz, because I cannot discern the difference.

Due to Nyquist-Shannon, the A/D converters actually shave off frequencies above half that of the sample rate. So 48kHz sees filters shave off 24kHz and up.

 

[jesussaddle]

Hey brother, I think I have the answer for you.

Your running ASIO4ALL with Fireface is definitely not recommended, but I can understand why you did it. The rule of thumb is...any interface that has it's own ASIO driver, you should always run that. I've never seen a case where ASIO4ALL worked better than the actual driver for an interface. Now, if it was a faulty driver...there is a possibility. But you should always stick with the manufacturer driver for anything ASIO unless instructed to do otherwise.

Yes, very informative!

I really don't ever use ASIO 4 ALL except that I tested it for cases involving a large number of tracks, when I was putting together my new PC.

The official final Fireface driver is the one I was using when I heard this difference. There is a good chance that, even though Ableton has very little choice in terms of Prefs, I am experiencing something basic in terms of a mixing process somewhere. I have my stereo tracks set normally, and I don't necessarily think pan law is the answer, but if Ableton is somehow creating a different stereo field, it could be. But my hunch goes with some others who to me personally have said it may be due to process unreachable and behind the scenes, that allow Ableton Live to do its excellent audio stretching. Is that possible?

 

[jesussaddle]

I just have to comment on this. Whether, you are sampling at 48 kHz or 96 kHz has nothing to do with, if you record pitches higher than 20 kHz. It has to do with the amount of harmonic content you pick up and the ability to regenerate the incoming waveform.

If you record a 20 kHz tone at 48 kHz, you get two sample per period (one full evolution of a sine wave). So when the digital signal has to be converted back to the analog domain, it can only assume that the incoming signal was a perfect sine wave, which may or may not be the case. At 96 kHz, the A/D converter picks up four samples (almost five) per period. This gives the D/A converter much more information to work with when regenerating the analog signal.

All that said, I still record at 48 kHz, because I cannot discern the difference.

This reminds me of my experiment conducted for a vinyl shop owner/family of viola/cello players so they could teach me about why vinyl is so good, lol (I'm taking liberties with their intent - they probably knew I didn't know what the $%@ I was doing.)

What they had me do was get a monophonic vinyl record and compare this with a stereo one, and a CD. Piano recording. (I did this with the help of a guy who won a Grammy for their remastering one of Disney's old classic musical soundtracks - so it was entirely unscientific and off the cuff, but I blindingly trusted this person to confirm what I heard).

When you live in a wood-floored house with hallways and rooms off to the side, and you have your speakers in the main living area (which had varieties of wood as well as a fireplace with some stone/brick), the mono vinyl recording travels all over the house. You can go into the rooms off to the side of the hall, and hear a clear piano recording.

So their basic point was that stereo recordings (minimization of this with different mic setups may also be a factor), at least when reproduced by speakers, were creating cancellation, some even at inaudible higher frequencies that impacted lower frequencies. Whatever the frequencies and inter-impact between them, this apparently reduced the liveliness of the physical result. When you produce an accurate representation of higher frequencies, isn't it possible that they bounce around and creating some positive cancellation? C'mon, it could be positive. You have to be open to this sort of thing. Okay, okay, cancellation and wave interactions are all bad, bad, bad, bad. Sadly then, no benefit ever comes from inaudible-frequency waveforms

 

[Smittefar]

Due to Nyquist-Shannon, the A/D converters actually shave off frequencies above half that of the sample rate. So 48kHz sees filters shave off 24kHz and up.

Yes - but our ears shave off anything above 20kHz, so that is not a problem - and not the reason to go to 96 kHz - 96 kHz is used for high resolution audio, and it does have twice the number of data points per cycle as 48 kHz, and thus it resolves the harmonic content better. The purpose is not to be able to record 30 kHz audio.

 

[Smittefar]

@jesussaddle: When they played the mono recording, did they only use one speaker? It is my understanding that a mono recording played from multiple speakers could just as easily create cancellations of different kinds.

You are right that all sorts of weird cancellations can occur, and i am no expert on cancellations. But I maintain that the main reason for 96k audio is better representation of harmonic content, i.e. the shape of the waveform.

 

[jesussaddle]

@jesussaddle: When they played the mono recording, did they only use one speaker? It is my understanding that a mono recording played from multiple speakers could just as easily create cancellations of different kinds.

You are right that all sorts of weird cancellations can occur, and i am no expert on cancellations. But I maintain that the main reason for 96k audio is better representation of harmonic content, i.e. the shape of the waveform.

We didn't get that advanced, so yeah, they could have created some amount, because we basically had the two speakers separated by about 10 feet, and pointed lengthwise into the living area (i.e. perpendicular).

But my question is, isn't wave interaction a double-edged sword? I mean, don't a lot of synthesizers use wave combination to create a lot of desirable results? And on the other hand, when a close similarity exists and the waves are mixed DIRECTLY this is almost always not good.

I know its not scientific, but one reason I always thought so was because I bought headphones once that were said to be capable of reproducing down to 4 hz and up to 24k. They were always my favorite pair, despite my having others that were at least as hi-fi, but just not said to cover that low range. Because of that I always just assumed that what I DIDN"T HEAR was still influencing what I did, in a positive way.

 

[Smittefar]

Our perception of sound is completely dependendant on compined waveforms - A sine wave is perhaps the most boring sound ever. To make it sound like anything, you have to add harmonics which will alter the waveform. The right amount of harmonics will give you sawtooth, another amount will give you square wave. Guitar sound is a completely different set of harmonics (the relation between the harmonics changes with time). Distortion is making the guitar sound clip, which in turn make the waveform look more like a square wave - so yes interactions are fantastic. The problem arises, when you have two or more waveforms that are close in frequency or in phase. This may cause all sorts of weird artefacts that are not pleasing.

 

[mr_fender]

This reminds me of my experiment conducted for a vinyl shop owner/family of viola/cello players so they could teach me about why vinyl is so good, lol (I'm taking liberties with their intent - they probably knew I didn't know what the $%@ I was doing.)

What they had me do was get a monophonic vinyl record and compare this with a stereo one, and a CD. Piano recording. (I did this with the help of a guy who won a Grammy for their remastering one of Disney's old classic musical soundtracks - so it was entirely unscientific and off the cuff, but I blindingly trusted this person to confirm what I heard).

When you live in a wood-floored house with hallways and rooms off to the side, and you have your speakers in the main living area (which had varieties of wood as well as a fireplace with some stone/brick), the mono vinyl recording travels all over the house. You can go into the rooms off to the side of the hall, and hear a clear piano recording.

So their basic point was that stereo recordings (minimization of this with different mic setups may also be a factor), at least when reproduced by speakers, were creating cancellation, some even at inaudible higher frequencies that impacted lower frequencies. Whatever the frequencies and inter-impact between them, this apparently reduced the liveliness of the physical result. When you produce an accurate representation of higher frequencies, isn't it possible that they bounce around and creating some positive cancellation? C'mon, it could be positive. You have to be open to this sort of thing. Okay, okay, cancellation and wave interactions are all bad, bad, bad, bad. Sadly then, no benefit ever comes from inaudible-frequency waveforms

That's not a very valid comparison. Stereo separation on vinyl is much worse than CD because of the way the two channels are carved into one record groove. One needle is "reading" both channels so there's cross-talk between them and the low end in vinyl recordings is often summed to mono as well. Stereo CD audio is two completely independent data streams, so there is no cross-talk at all. Because of this, LP content and CD content are usually mixed and mastered differently, so you are not comparing apples to apples.

 

[dr bonkers]

That's not a very valid comparison. Stereo separation on vinyl is much worse than CD because of the way the two channels are carved into one record groove. One needle is "reading" both channels so there's cross-talk between them and the low end in vinyl recordings is often summed to mono as well. Stereo CD audio is two completely independent data streams, so there is no cross-talk at all. Because of this, LP content and CD content are usually mixed and mastered differently, so you are not comparing apples to apples.

plus on vinyl, everything below 100 Hz or so is usually summed to mono and aggressively limited as part of the mastering process for vinyl. Without knowing that detail, juesussaddle's test experience could have been even more compromised.

 

[jesussaddle]

plus on vinyl, everything below 100 Hz or so is usually summed to mono and aggressively limited as part of the mastering process for vinyl. Without knowing that detail, juesussaddle's test experience could have been even more compromised.

True. In fact, I didn't even have a CD of the same music. Its all highly suspect, but I s'pose its still worth doing if you're completely beside yourself with time, resources, and Phillip K. Dick loving females. Trust me, when you're in a room, around a corner, and you hear highs like those upper keys on a piano, sounding clear and defined, its a surprise. Mono can cause that if untreated. Reproducing it on CD, or with stereo, I suppose if that happens then I personally will adjust my theories, but until then, I have a personal experience, albeit completely unacceptable as a minority report thus far. Moreover, Philip Dick's pic of the future is completely unreliable also, as were his tastes in women if recollections serve. So what if it has vague, coincidental accuracies, and if his fans, and those of H.P. Lovecraft, make great lesbian couples. In reality we will never know what its like to hear an android's thoughts or H.P. Lovecraft read from the bathtub.

Except that last one.



(Sorry, can't wait to be back on topic once the FW comes.)

 

[vinnieRice]

The difference between 44.1/16 and 48/24 is huge and everybody should be able to hear that. The difference between 48 and 96 is much more subtle and in a blind testing situation would be hard to determine. However... that particular situation is irrelevant. Put up a DAW mix with dozens of tracks, some nice analogue eq and comp inserts (with all the inevitable AD conversions), and play around with that for a few hours and it just sounds better as a 96 session than at 48 - for whatever reason. The nicest my AFX has ever sounded is from the analogue outs through a pair of Neve 1084's recorded at 96. Absolutely glorious. Hollywood soundtracks are recorded at at least 192kHz - makes a difference when you are mixing 100+ orchestral tracks. TC Electronic run their DACs at 44.1 - Strymon at 96, make of that what you will.