Way for logic to treat session sample rate change like Ableton does?

[djohnq]

I might be wrong about what Ableton is doing in my description below, but as I understand it, I think its doing what I'm saying. I just started attempting to work this way so I doth have any real world experience of doing this with Ableton.

 

In Ableton, you can work on a session in 96KHz and record audio files at 96KHz, then if the session gets too nasty cpu wise you can change the sample rate down to 48KHz and keep working on it. Then when Its all done you can switch back to 96KHz for bouncing and stem creation etc. The actual audio files in the session remain at 96KHz the whole time, meaning Ableton is doing some sort of real time sample rate conversion on all the audio files

 

In Logic, it would appear that if you want to switch from 96 to 48, you would actually have to make copies of the audio files at the new sample rate. This would mean if you are doing something like flex pitch, or even if you are just fading or editing the regions, all that work would be lost when switching back to 96KHz because you would have to re import your 96KHz audio files, or if you wanted to keep all those changes you made at 48 intact, you would have to convert those 48KHz audio files back to 96KHz which defeats the purpose in the first place.

 

Is there a workaround or a way to do in Logic what Ableton is able to do regarding this?

 

Thanks!

[des99]

I don't know the answer to your question offhand, but what is the reason you want to record at 96KHz, out of interest?

[djohnq]

I don't know the answer to your question offhand, but what is the reason you want to record at 96KHz, out of interest?

 

Well I have been going back and forth on this. I guess the short answer would be because it is better quality. I'm no expert mix engineer but I am working to get better. I THINK I can hear the difference in sample rates but I might be fooling myself. I think the real difference in all the sample rates is to watch out for what its doing in terms of aliasing. But regardless of all that, if I am working with someone who can actually tell the difference (or believes they can), then I feel its best to just work at the highest feasible sample rate to keep everyone happy).

I'd still be interested in the answer of my orignal question, but hey, if anyone has links to top tier mix engineers or producers talking about sample rate throughout the whole process (tracking, mixing, oversampling in plugins, mastering, etc.) that would be awesome to see too! No offense to anyone but posting their personal opinion about sample rate is absolutely meaningless since we don't know if you took an audio class online that had one video about the subject or if you've worked on 100 songs that were nominated for Grammys.

 

By the way, here is a great video about the actual science of it.

 

 

So yeah I mean the long and short of it is that it's complicated. I am more interested in hearing what the most popular sample rate is in the world of top 40 pop type stuff. I'm sure there is a different answer about ancient rare flutes from the end of the earth that have intricate overtones and complicated dynamic ranges that need to be preserved. But Im talking about straight fire hits haha.

 

Like if a producer is starting up a track and having an artist come in to record vocals on it. What sample rate are they running that session at? Are they maybe running their VST at a certain rate while making the track and then bouncing and changing the sample rate for the tracking portion?

 

I've "heard" from reading and asking around forums that 48KHz is the go to sample rate these days for pop sessions but I am struggling to find a video or interview of an actual known hitmaker talking about this subject. The results instead It seems littered with opinions and people getting on their high horse acting like they are audio geniuses about the subject which makes them less believable in the end.

 

I guess until I get the relief of seeing someone with a lot of credibility say "yeah, I do all my producing and mixing and the whole chain of making the song at 48KHz and so do all of my big shot friends" then I am going to stick to keeping things at 96KHz.

[robinloops]

If you really want to work that way, you could covert the files mid project but first make a backup of the original sample rate files. Then work at 48 until the end then convert again but replace the twice converted audio files with the original sample rate backup copies.

 

Seems like a convoluted workflow though. I’d suggest to either work in 48 the whole way or 96 the whole way through.

[djohnq]

If you really want to work that way, you could covert the files mid project but first make a backup of the original sample rate files. Then work at 48 until the end then convert again but replace the twice converted audio files with the original sample rate backup copies.

 

Seems like a convoluted workflow though. I’d suggest to either work in 48 the whole way or 96 the whole way through.

 

 

Nice thanks! I wouldn't mind that except that makes the audio off limits for editing while in 48KHz as far as i can tell?

 

Like if I have a session with a bunch of vocal tracks, and I change the sample rate to 48 and do that whole copy/replace function in the audio bay thing to convert the audio files to 48KHz as well, then I start editing fades or maybe cut up the clip or something or edit the waveform while in 48K, Im thinking all of those edits would be lost when replacing the original file back to 96KHz when I want to start boning tracks, mixing, and so on. However, maybe there is a workaround for that so Ill look into it. Like maybe you can replace the file and keep all fades and cuts?

 

To clarify my post, what I do now is simply freeze the cpu offending tracks when resources get low. But it would be nice if there was an was a way to not have to freeze everything and just have the VSTs and plugins temporarily in 48KHz and the audio read as 48KHz but having the file retain all the 96KHz info.

Edited June 11, 2020 by djohnq

[des99]

I guess the short answer would be because it is better quality.

 

Well I was hoping you weren't going to say something like that and you were going to list a specific reason for doing it.

 

If your reason is just "well, the numbers are bigger so surely it's better", then I'd strongly argue to stick to 44.1 or 48KHz.

 

I think the real difference in all the sample rates is to watch out for what its doing in terms of aliasing.

 

Yes - there *can* be improvements here, but a lot of plugins where this is an issue oversample anyway (so they internally work at multiples of the sample rate to get better aliasing performance.) so you don't need to run your entire system at 400KHz just to get those benefits for the processes that require them.

 

But regardless of all that, if I am working with someone who can actually tell the difference (or believes they can), then I feel its best to just work at the highest feasible sample rate to keep everyone happy).

 

Again, this is a "I don't understand what's going on, so I'll stick to a recipe in the hope it's ok." If you don't know you *need* high sample rates, I strongly recommend against using them. It's a waste of resources for very little gain, and if you're recording audio, unless you are working in super-high end studios and gear you won't get any practical differences - and are just chewing up disk and processing resources pointlessly (imo) recording frequencies your input chain is unable to deliver.

 

I'd still be interested in the answer of my orignal question, but hey, if anyone has links to top tier mix engineers or producers talking about sample rate throughout the whole process (tracking, mixing, oversampling in plugins, mastering, etc.) that would be awesome to see too! No offense to anyone but posting their personal opinion about sample rate is absolutely meaningless since we don't know if you took an audio class online that had one video about the subject or if you've worked on 100 songs that were nominated for Grammys.

 

You may not know much about posters giving you feedback, but many have us have been around for *years* and have demonstrated for years we kinda know what we are talking about. And there are plenty of known, proven track record guys around in these kinds of places.

 

My personal opinion is strongly linked in an understanding of the facts, and is linked to my experiences, and of course, not everyone will agree with me. Some of those will be for good reasons (they get appreciably better results working at higher sample rates, or need them for particular reasons), but some will be like they don't really understand what's going on, how convertors work, what the trade-offs are and so on, and have this notion that bigger numbers must always be better.

 

But yes, I do recommend that if you're interested, certainly have a broad read around on the subject.

An informed decision is always better than an uninformed one, or a guess...

 

I am more interested in hearing what the most popular sample rate is in the world of top 40 pop type stuff.

 

Well that's easy, and it's 44.1KHz.

 

Like if a producer is starting up a track and having an artist come in to record vocals on it. What sample rate are they running that session at? Are they maybe running their VST at a certain rate while making the track and then bouncing and changing the sample rate for the tracking portion?

 

Most of the time, 44.1KHz, and if it's for a video session, the standard is 48KHz. 48KHz also has some benefits in that the reconstruction filters don't need to be quite so steep, but for most uses that difference is likely to be tiny anyway.

 

I've "heard" from reading and asking around forums that 48KHz is the go to sample rate these days for pop sessions but I am struggling to find a video or interview of an actual known hitmaker talking about this subject. The results instead It seems littered with opinions and people getting on their high horse acting like they are audio geniuses about the subject which makes them less believable in the end.

 

There are always people who understand this stuff from solid engineering principles, and there are always people who rigidly stick to what works for them without understanding why. There are also people that hear something on a forum and blindly take that to be true too. I'm suggesting that taking blind advice from *anybody* is probably not a great idea, unless you know damn well they know what they are talking about (which can of course be difficult to identify if you don't have decades of experience etc.), but look into it youself, and maybe do some tests too if you're that way inclined.

 

I guess until I get the relief of seeing someone with a lot of credibility say "yeah, I do all my producing and mixing and the whole chain of making the song at 48KHz and so do all of my big shot friends" then I am going to stick to keeping things at 96KHz.

 

Hey, it's your computing resources to waste , and if your machine is powerful enough and that's what you want to do, by all means go for it without understanding why. Just understand that actually many convertors have a slightly worse performance at higher sample rates for various technical reasons and you certainly don't need them to reproduce the range of human hearing "better", as 44.1KHz can perfectly represent the range of things we can possibly hear, and a 24-bit dynamic range is also way more than we need and more than convertors are able to achieve.

 

If you think your microphone is generating meaningful audio data at 40KHz+ on a vocal session and your song will suffer because of it - well, it's fairly easy to test and prove whether that's the case (hint: I know the answer to that one already!

 

If you want to be sure, 24-bit/48KHz will be plenty, and nobody you are going to work with is going to have a problem with it. Except perhaps the other people who just think bigger numbers are better.

[David Nahmani]

58kHz? 98kHz? I suppose those are typos. In any case I would strongly recommend you simplify your workflow and always work at 48kHz. All the hit songs you hear on the radio? 48kHz. I know this from discussions with top 40 producers and chief engineers in some of the biggest recording studios. Last time I worked at the Record Plant in L.A. I asked their chief engineer what sample rates their clients were typically using. He said 99.5% of the sessions run at 48kHz.

 

Oh and unless you've spent several thousands dollars in a professional clock to clock your top of the line A/D converters, you're getting less jitter at 48kHz, so a better quality than when recording at 96kHz.

[des99]

Oh and unless you've spent several thousands dollars in a professional clock to clock your top of the line A/D converters, you're getting less jitter at 48kHz, so a better quality than when recording at 96kHz.

 

Indeed. Some tech specs like jitter drop as you crank the sample rates (obviously, newer gear is a bit better in this regard) and as I said above, everything is a tradeoff.

 

Basically, you're lowering your quality just to be able to record higher frequencies that won't even be there anyway, with the fact that you're halving the performance of your system just to reproduce all the stuff that isn't there. To *me*, that's not an acceptable tradeoff, which is why I don't use high sample rates for 99.9% of the stuff I do.

[djohnq]

I guess the short answer would be because it is better quality.

 

Well I was hoping you weren't going to say something like that and you were going to list a specific reason for doing it.

.....

 

Nice thanks for all the info, definitely helpful!

 

Reading back my post I see I sound pretty dismissive about the validity of replies on forums, which is obviously incorrect considering the amount of knowledge I gained from gearslutz and DUC etc, and surely here as well now that Im using Logic. And also to get meta about it Im in a forum asking a question so my whole attitude about forum opinions/facts doesn't make sense haha.

 

Really I was just trying to say why I was looking to see if anyone had any links to interviews or videos about the subject.

 

Anyway thanks fro the advice!

[djohnq]

58kHz? 98kHz? I suppose those are typos. In any case I would strongly recommend you simplify your workflow and always work at 48kHz. All the hit songs you hear on the radio? 48kHz. I know this from discussions with top 40 producers and chief engineers in some of the biggest recording studios. Last time I worked at the Record Plant in L.A. I asked their chief engineer what sample rates their clients were typically using. He said 99.5% of the sessions run at 48kHz.

 

Oh and unless you've spent several thousands dollars in a professional clock to clock your top of the line A/D converters, you're getting less jitter at 48kHz, so a better quality than when recording at 96kHz.

 

Okay cool nice thanks for the info! Yeah in terms of the actual quality of 96 vs 48, I guess I do care about that to a degree, but what I am really interested in more is what the industry standard is. So that is really helpful to hear more affirmation that it is 48 currently. Sounds like it used to be 44.1, then it morphed to 48 at some point.

 

I guess in my head is like Im wondering like "well is it starting to move to 96 or going to any time soon?" Sounds like its not which is great to know about!

 

Im sure one of those new Mac Pros could run a pretty dang impressive session at 96KHz so maybe some producers with top of the line A/D converters start working at 96? In which case, in the astronomically small chance that I one day end up working with someone who does have that top notch gear, or the even smaller chance that I own high end gear like that myself one day, then 96KHz would make sense.

 

Also thanks for this site David, I just started using Logic a few weeks ago and my workflow speed is already up to par with my last DAW workflow speed and its mostly because of all the answers I have found on here!

 

Which reminds me (I knew I shouldn't have taken Des99's question bait haha) I am still going to look for a workaround on the original question of this thread and Ill post back if I find anything that makes sense. Regardless of whether I should be working at 48 or 96, I am still interested to know if there is a way to run a "XX"KHz session with "XY"KHz audio files in the session without needing to create new files. Im pretty sure the answer is no though

[des99]

Sounds like it used to be 44.1, then it morphed to 48 at some point.

 

There was a time when most (non-video) production audio was destined for CD, so it made sense to stick to 44.1 to avoid the need for sample rate converting, which was a destructive process and at the time it didn't sound too good.

 

Nowadays, though we have better SRC algorithms, CD is not nearly as dominant a music product (most ends up in digital formats, mp3/flac/etc) so you can get the slight benefits of 48KHz in general, and if need be you can SRC to 44.1KHz decently enough if necessary.

[djohnq]

Okay so I did a whole bunch of tests where I performed region cuts, fades, and flex pitch modifications on an audio file in 96. Then I switched the session to 48, did the copy/convert on the audio file to 48 in the project tab, and then tried a whole manner of techniques to get the flex pitch to adapt to the new sample rate (with the idea that I would do a file replace of the original 96 version of the file when going back to 96 and figure out how to update the flex at that point as well). It was all a bunch of complicated maneuvers and got confusing, every one in a while I would think I figured it out, and I may have, but I don't think I did. Point being the whole process is too complicated like robin loops suggested above.

 

If someone knows if there is a way to :

start session in 96 -> make flex pitch edits on an audio region -> switch session to 48 -> replace the 96 file with a 48 version of the file and have the flex pitch adapt correctly -> make flex pitch edits in 48 -> switch session back to 96 - replace the 48 file with the original 96 file and have the flex pitch adapt correctly

That would be cool! But I still dont think its possible

 

So, as much as I was starting to like flex pitch, standard Melodyne (non ARA) will allow the session rate to change and playback at 48 while keeping its transferred file at 96. ARA might allow that as well but I dont use it, seems a little buggy, but more importantly, it won't allow you to change the project tempo and re-sync correctly like the non-ARA plugin version does.

 

Of course, using the Melodyne method only works correctly if all audio tracks in your session are melodyned, otherwise anything that is not melodyned would need to be dealt with accordingly. I haven't looked in to that too much yet, basically doing testing of what I described above but not involving flex pitch (just doing cuts and fades). I might look into it, but the solution will be pretty nasty no doubt if is possible haha. Obviously Audiosuite type processing would be off the table on those files as well when in 48KHz mode.

 

The whole reason I want to do this is so that I can work at low cpu when the session gets big, but when everything is done and Im ready to commit the Melodyne to an audio track, I can switch back to 96KHz and get all the good processing resolution.

 

And then back to the whole tangent of this thread: Everything I am talking about above is a perfect example of why 96 would be a good idea. Wouldn't Melodyne and Flex pitch both be great reasons to record at 96? Im stretching stuff in Melodyne all the time. Surely the 96KHz file is better to stretch right?

[triplets]

If you're gonna be doing a lot of flexing and stretching, higher sample rates are better.

Changing sample rates within a session or converting files back and forth is overcomplicated and not gonna give you any advantages.

Stick with one sample rate and that will be a better workflow.

My 2 cents.

[des99]

I can switch back to 96KHz and get all the good processing resolution.

 

96 is not giving you any better "resolution". 44KHz can reproduce *perfectly* audio frequencies up to 20KHz - the limit of the very best of human hearing. This is proven well by math and DSP.

 

"Resolution" (which generally comes from digital graphics) actually doesn't have an equal concept in audio - the closest it comes to is word length, which, at 24bit is more dynamic range than we can hear and that convertors can properly handle anyway.

 

All you are really getting with 96KHz is the ability to work with audio frequencies higher than 20-22KHz. 96KHz does not somehow make the recorded 20-20KHz human frequency range "more precise", "better resolution" or "higher quality' in any way...

 

I'm just trying to correct some of these misconceptions.

 

And your workflow looks appalling just to let you store the audio content in the 20-40KHz range(!)

(Have you even *tested* how much actual audio content/energy you have in this range on your 96KHz recorded files?)

 

Anyway - I'm not going to tell you not to do this - it's your choice of course - but I just hope you come to understand better the fundamentals of digital audio and make informed decisions, rather than just follow your intuition as to how you think this stuff works, because digital audio is often surprisingly different in reality to how you might think it works. There are *still* people around who think that digital audio is stair-stepped, and smaller steps means higher precision... It is difficult to dispel some of these long-standing digital myths, and I'm making a point of it here not necessarily for the OP, who seems to have his mind made up, but for others who maybe stumble upon this topic in the future.

[robinloops]

Are you actually getting system performance issues with the higher sample rate? If so, Why not just work in 96 and change your buffer if you need a little boost in performance when the project gets large? And use low latency mode with a smaller buffer if you need to record new parts? Then you wouldn't have to worry about convoluted workflows changing sample rate mid project...

[djohnq]

and I'm making a point of it here not necessarily for the OP, who seems to have his mind made up, but for others who maybe stumble upon this topic in the future.

I guess I am doing a bit of the same by responding here.

 

I think the "perfect representation" you are talking about is all about the fact that if you record a file at 48 KHz, it will be able to playback everything the same perceived quality as 96 file because the human hearing only goes up so high. so anything we hear can be output at 48 and when it is converted to analog the analog conversion has enough info to redraw perfectly what the human ear is able to perceive. And Im sure Im not explaining it exactly correct but thats the basics of your argument, right?

 

But im not talking about that. Im talking about degradation of sound quality when processing is introduced within the session. If you are stretching audio, all of the sudden those extra data points in the 96KHz file make a difference because they are being used to help recreate the new waveform you are making when you are stretching. And there is more data to work with at the higher sample rates therefore different results (but maybe not always better, see below). I time stretch audio all the time so to me this is an important thing to consider.

 

So I ran an experiment.

 

For starters, without even doing an experiment, I should stop and point out that its not even necessary to go that far. To set up for the experiment, I loaded Xfer Serum and made a little 2 bar sequence with patch (SQ DYK PYK). The session was in 44.1. I played the sequence I made and said okay that will do for the experiment. Then I switched the session to 96KHz. And oh geez! The reverb in serum was completely different. So much cleaner. I mean I always assumed running my synths at 96KHz was better but never really investigated it too much before this whole thread got me thinking hard about it. So just that alone. Just do that to see for yourself. Its so evident how much the higher sample rate makes a difference. Like you don't need to go to audio school to hear the difference. Its obvious. And just to be sure, I bounced both the 48 session and the 96 session to a 48KHz file. And the 96 session bounced to a 48KHz file still sounded way better than the 48KHz session bounced to 48KHz file. I knew that would be the case but wanted to be thorough to save anyone any time by posting asking if I did that, haha cause I did. Also I went into oscillator setting in serum and set the oscillator to 4x while the session was running at 48. It still sounded worse that the 96 session rate with 2x oscillator in serum.

 

So back to the experiment, I couldnt even use that patch for the experiment because the sounds were so different. I used a different sound that sounded pretty similar between the two sample rates. I will admit, the experiments had some surprising results, but for the reasons stated above the experiment doesn't even matter anymore. It is already clear as to why it is a good thing to run a session at 96KHZ if you can and you run AU synths. Running an AU synth in a session is a specific need, but Id have to think it is a pretty common thing to do these days. So onto the experiments:

 

This experiment was set up to be super extreme so I could hear the differences. I put Xfer Serum AU and made a quick little region using a sequenced patch to make a short clip with complex sounds . I set the session to 96 and bounced in place the Serum track. On that 96KHz file I created, I set its track to flex pitch mode. And I time stretched it a very exaggerated amount. then I turned that track off. I set the session sample rate to 48. I bounced the serum AU track again to make a 48KHz version of the bounce. I set it to flex pitch and time stretched by the same exaggerated amount. Then I listened to that track. Then I set the session sample rate to 96 and listened to the 96 version. I went back and forth. To me, the 96KHZ version sounded better. There was one part of the file in the 48 version that had a really harsh distortion on a certain note. I went back and listened to that same note in the 96KHz version and the distortion was not as bad. That being said, that was just a quick test I set up now so I obviously might be missing important factor that I am not considering.

 

That all said, they both sounded degraded obviously so it really is down to a taste at some point with some of that. The experiment was intended to be extreme so I could hear differences easily. I messed around with the experiment above with Melodyne instead of flex pitch as well and will admit the sometime the results at 96 sounded worse than 48 to my ear. But that doesn't really matter anyway, The point is that there are complex processes going on in the processing chain and the sample rate makes a huge difference in the sound. So you can't just dismiss and say there is no reason to use 96KHz ever. In my experiments above, the differences are always notable. So maybe for each specific plugin or process it is worth exploring how the sound is affected. I might end up doing that one day, got really in depth on Melodyne, and post about what I find here.

 

I mean maybe I will end up recording vocals at 48, but the only reason would be if I feel like Melodyne works better at 48. Then , I still would be tracking vocals in 48 and melodyning them, then putting the session back into 96 eventually for the synths. Bottom line is I like to run my synths at 96KHz, and I I just discovered, I am very glad I have been doing that cause to me it sounds way better. So for me, I like 96 KHz sessions. So we will see if that is enough of an explanation for Des 99 haha.

 

Most of the time I can run the session at 96KHz an not have a problem. But every once in a while a session will get too heavy, which is why I started this thread. to find out if there is a workaround in the rare time I need to free up some CPU. Sure I can freeze tracks, but if there is workable solution to run the session in 48 whole keeping the audio files at 96, then to me that is the most flexible and easiest workflow. Really doing both freezing and samples rate conversion is what I would end up doing depenindin on what im doing in the session at the moment. To me, if I can work in 96 most of the time, and only every once in a while I have cpu problems, why would I switch to always doing 48 if it means my synths will have worse quality and there still is a possibility that there is a workaround in Logic?

 

As for my workflow, I always like to say "to each their own!". Anyway, in response to this:

 

And your workflow looks appalling just to let you store the audio content in the 20-40KHz range(!)

(Have you even *tested* how much actual audio content/energy you have in this range on your 96KHz recorded files?)

 

Im guessing the appalling part is in response to:

 

start session in 96 -> make flex pitch edits on an audio region -> switch session to 48 -> replace the 96 file with a 48 version of the file and have the flex pitch adapt correctly -> make flex pitch edits in 48 -> switch session back to 96 - replace the 48 file with the original 96 file and have the flex pitch adapt correctly

 

Well, I mean, if it is doable, then once the process for doing it is defined, Im sure a little elbow grease, hot keys, and macros between Logic and Finder would make this task very simple to perform on hundreds of files at a time if needed. And its something I would only do once or twice per session so even if it took like a minute or so it would be worth it for me.

[djohnq]

Found a video asking Warren Huart about sample rates higher than 48:

 

Edited June 14, 2020 by djohnq

[djohnq]

Here is George Massenburg talking about 96KHz:

 

[djohnq]

Here is Al Schmidt talking about 96KHz and 192:

 

[djohnq]

Here is Tom Elmhirst mentioning that he does 24/96 mixes:

 

[djohnq]

part 2 of George Massenburg talking about when he considered digital to be "good digital"when he started doing session in 96KHz

 

[djohnq]

Here is a video of Streaky talking about a bunch of different sample rates throughout the whole process. The whole video is good, talks about the mix vs mastering and sample rates.