Q: Oversampling - How does Reason & ASIO4ALL handle it?

Have an urge to learn, or a calling to teach? Want to share some useful Youtube videos? Do it here!
Post Reply
RobC
Posts: 1833
Joined: 10 Mar 2018

09 Jun 2022

I couldn't find the answer...

If Reason is run at a higher sample rate than the Audio Interface, using ASIO4ALL, then is there any digital filter applied by either, before the digital audio is sent to the Audio Interface?

I'm sure that resampling does happen, otherwise it wouldn't sound right. If that happens, filtering would make sense to prevent aliasing.

How much does this alter the sound, and how CPU-heavy is the process? Or is it better to just run at the Audio Interface's native sample rate (and get one with higher sample rate support if I want that)?

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

09 Jun 2022

I don’t know about ASIO specifically, but on a Mac you cannot set the Reason sample rate - you set the INTERFACE sample rate and Reason follows. On some interfaces, like Focusrite, you can change the sample rate from the Reason dialog but you are changing the FOCUSRITE sample rate in this case NOT just the Reason sample rate which always follows the interface rate.

So I would first have to ask why would ASIO allow a different sample rate for software vs hardware? What is the use case? Is it REALLY doing this, or does it work like the Focusrite interface on Mac?
If it IS doing this, then you most definitely need proper sample rate conversion involved and that certainly should involve a filter among many other things…
Selig Audio, LLC

User avatar
jam-s
Posts: 3035
Joined: 17 Apr 2015
Location: Aachen, Germany
Contact:

09 Jun 2022

Re-sampling (which always needs a filtering step at the end) is usually the task of the driver of the audio interface. So it can be done either in software or in hardware depending on the capabilities of the hardware.

According to the asio4all FAQ you can chose if asio4all shall do the re-sampling in software:
Why does it sound muddy/why do I hear crackles in intervals of like 10 seconds or so when running at 44.1kHz where at 48kHz everything is fine?

These are typical resampling artifacts from poorly designed WDM drivers. Since version 1.7 beta 2, there is a check box ‘Always Resample @44.1k’. When checked, ASIO4ALL will take care of the resampling business on its own, even if the card driver says it can do 44.1k – Alles wird galaktisch gut!

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

09 Jun 2022

RobC wrote:
09 Jun 2022
1-How much does this alter the sound, and
2-how CPU-heavy is the process?
3-Or is it better to just run at the Audio Interface's native sample rate (and get one with higher sample rate support if I want that)?
1-Bottom line, how does it sound to you? What is your end goal? Only you can determine how much it is changing the sound, since you’re the only one who can hear the results!
2-How much of a CPU hit does it take on YOUR system? That is how CPU heavy the process is… ;)
3-IMO it’s always better to monitor at the same sample rate you are working at. Otherwise you are potentially listening to things that will not necessarily be a part of the final output/product. That’s the main reason I’ve always worked at the final output sample rate, either 44.1 or 48 kHz (for video).

[realized I didn’t actually answer any of your specific questions with my initial response]
Selig Audio, LLC

User avatar
orthodox
RE Developer
Posts: 2286
Joined: 22 Jan 2015
Location: 55°09'24.5"N 37°27'41.4"E

09 Jun 2022

RobC wrote:
09 Jun 2022
I'm sure that resampling does happen, otherwise it wouldn't sound right. If that happens, filtering would make sense to prevent aliasing.
Resampling *is* filtering, it's a single indivisible process.
RobC wrote:
09 Jun 2022
1) How much does this alter the sound?
2) How CPU-heavy is the process?
1) A good resampler alters the sound by just -140dB (apart from filtering out the HF content that can't be represented in the lower sample rate when downsampling).
2) It's heavier than EQ, but lighter than FFT.

User avatar
moofi
Posts: 1024
Joined: 19 Jan 2015
Location: hear

09 Jun 2022

140dB difference sounds like almost nothing :-D

orthodox wrote:
09 Jun 2022

1) A good resampler alters the sound by just -140dB (apart from filtering out the HF content that can't be represented in the lower sample rate when downsampling).
2) It's heavier than EQ, but lighter than FFT.

User avatar
orthodox
RE Developer
Posts: 2286
Joined: 22 Jan 2015
Location: 55°09'24.5"N 37°27'41.4"E

09 Jun 2022

moofi wrote:
09 Jun 2022
140dB difference sounds like almost nothing :-D
:D
I mean, by adding noise at -140dB. The actual difference for normalized signal will be 0.000001dB.

RobC
Posts: 1833
Joined: 10 Mar 2018

10 Jun 2022

selig wrote:
09 Jun 2022
I don’t know about ASIO specifically, but on a Mac you cannot set the Reason sample rate - you set the INTERFACE sample rate and Reason follows. On some interfaces, like Focusrite, you can change the sample rate from the Reason dialog but you are changing the FOCUSRITE sample rate in this case NOT just the Reason sample rate which always follows the interface rate.

So I would first have to ask why would ASIO allow a different sample rate for software vs hardware? What is the use case? Is it REALLY doing this, or does it work like the Focusrite interface on Mac?
If it IS doing this, then you most definitely need proper sample rate conversion involved and that certainly should involve a filter among many other things…
Hmm, I'll check with my USB microphone... Okay, I remembered correctly, ASIO4ALL lets me set the sample rate in Reason independently up to 192 kHz, even though the USB microphone's DAC supports 16 bit / 48 kHz tops. While my dying main DAC is recognized as 32 bit / 192 kHz. ASIO4ALL handles both of them (even the on-board integrated soundcard is available) together with Reason perfectly, running at the same time.

I guess it's driver dependent then in case of Windows. It seems that ASIO4ALL is doing a pretty good job.

As for why it allows all this? Well, live oversampling can come in handy. For some reason, I noticed that Reason's sequencer reacted better to automation in some cases. And that compressors seemed to have a better attack. I think pitch modulation, such as FM sounded better. The higher 192 kHz sample rate was useful for detailed/surgical editing when it comes to samples (sound design). Oh, and obviously taking care of unwanted aliasing.

Now, if it's really working? I'll test that today with a 96000 Hz sine tone. If there's no aliasing sound, chances are it's getting resampled. I'm pretty sure that Reason does work at 192 kHz internally at least.
jam-s wrote:
09 Jun 2022
Re-sampling (which always needs a filtering step at the end) is usually the task of the driver of the audio interface. So it can be done either in software or in hardware depending on the capabilities of the hardware.

According to the asio4all FAQ you can chose if asio4all shall do the re-sampling in software:
Why does it sound muddy/why do I hear crackles in intervals of like 10 seconds or so when running at 44.1kHz where at 48kHz everything is fine?

These are typical resampling artifacts from poorly designed WDM drivers. Since version 1.7 beta 2, there is a check box ‘Always Resample @44.1k’. When checked, ASIO4ALL will take care of the resampling business on its own, even if the card driver says it can do 44.1k – Alles wird galaktisch gut!
I didn't use that 44/48 resampling, although I see it's there. So, something happens somewhere, cause it seems to work. But I'll do more testing today.
selig wrote:
09 Jun 2022
RobC wrote:
09 Jun 2022
1-How much does this alter the sound, and
2-how CPU-heavy is the process?
3-Or is it better to just run at the Audio Interface's native sample rate (and get one with higher sample rate support if I want that)?
1-Bottom line, how does it sound to you? What is your end goal? Only you can determine how much it is changing the sound, since you’re the only one who can hear the results!
2-How much of a CPU hit does it take on YOUR system? That is how CPU heavy the process is… ;)
3-IMO it’s always better to monitor at the same sample rate you are working at. Otherwise you are potentially listening to things that will not necessarily be a part of the final output/product. That’s the main reason I’ve always worked at the final output sample rate, either 44.1 or 48 kHz (for video).

[realized I didn’t actually answer any of your specific questions with my initial response]
It seemed transparent sounding. But in the past, you and others recommended not to use low pass filters on mixes for example, since it apparently can alter lower frequencies. That's why I worried, that maybe a low pass filter is added, but a later comment revealed that that's not exactly the case.
Honestly, I didn't really notice much of a difference on CPU use.
While I found some uses for higher sample rates, I'll probably consider a supported sample rate (by my future Audio Interface), too.
orthodox wrote:
09 Jun 2022
RobC wrote:
09 Jun 2022
I'm sure that resampling does happen, otherwise it wouldn't sound right. If that happens, filtering would make sense to prevent aliasing.
Resampling *is* filtering, it's a single indivisible process.
RobC wrote:
09 Jun 2022
1) How much does this alter the sound?
2) How CPU-heavy is the process?
1) A good resampler alters the sound by just -140dB (apart from filtering out the HF content that can't be represented in the lower sample rate when downsampling).
2) It's heavier than EQ, but lighter than FFT.
Okay, I definitely didn't know that! Thank you for clearing that up!
That's basically no difference for the human ear, no wonder it sounds transparent.
The CPU didn't show much difference, so I guess it doesn't drain that much.

User avatar
jam-s
Posts: 3035
Joined: 17 Apr 2015
Location: Aachen, Germany
Contact:

13 Jun 2022

RobC wrote:
10 Jun 2022
As for why it allows all this? Well, live oversampling can come in handy. For some reason, I noticed that Reason's sequencer reacted better to automation in some cases. And that compressors seemed to have a better attack. I think pitch modulation, such as FM sounded better. The higher 192 kHz sample rate was useful for detailed/surgical editing when it comes to samples (sound design). Oh, and obviously taking care of unwanted aliasing.

Now, if it's really working? I'll test that today with a 96000 Hz sine tone. If there's no aliasing sound, chances are it's getting resampled. I'm pretty sure that Reason does work at 192 kHz internally at least.
As CV and quite a lot of other internal Reason processing runs at 1/64 of the audio sample rate the effects you're observing seem to be the result of the faster internal CV processing for higher sample rates.

RobC
Posts: 1833
Joined: 10 Mar 2018

14 Jun 2022

jam-s wrote:
13 Jun 2022
As CV and quite a lot of other internal Reason processing runs at 1/64 of the audio sample rate the effects you're observing seem to be the result of the faster internal CV processing for higher sample rates.
Yeah, so if it will work without problems, I will stick to 192 kHz. Otherwise the 96 kHz will do, too.

I'm still not sure if there's any point to record vocals at 96 kHz in the 20-20kHz range. If the more detailed curves would make any difference for compressors. Though then again, if I set Reason to 96 kHz for example, at least I won't have to wait until it resamples everything.

Then again, Reason's original vocoder screws up (or at least used to) in FFT mode above 44.1 kHz, so hmm... more things to consider.

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

14 Jun 2022

“ But in the past, you and others recommended not to use low pass filters on mixes for example, since it apparently can alter lower frequencies.”

No, I never said that.
For every slight possible advantage you list for higher sample rates (which in some cases are minimal) there are also disadvantages which you don’t list (which in some cases are substantial).
Time and again I’ve found working at the final delivery sample rate is best.
Selig Audio, LLC

RobC
Posts: 1833
Joined: 10 Mar 2018

14 Jun 2022

selig wrote:
14 Jun 2022
“ But in the past, you and others recommended not to use low pass filters on mixes for example, since it apparently can alter lower frequencies.”

No, I never said that.
For every slight possible advantage you list for higher sample rates (which in some cases are minimal) there are also disadvantages which you don’t list (which in some cases are substantial).
Time and again I’ve found working at the final delivery sample rate is best.
Sorry, maybe I confused people ~ I had all kinds of threads. This topic discussed it, maybe I misunderstood something: viewtopic.php?f=5&t=7514064&hilit=filter
I remember I was once told not to filter a mix, because it would wreck it.

Well, I do know that higher sample rates will drain the system more, and come with bigger file sizes. And that down sampling, filtering will be needed. Maybe some plugins aren't compatible.

As for advantages, if we take the example with 1 wave cycle, or just a 1-shot synth sample that I spread out on the keyboard with a sampler, etc. - is it not worth it for that?
Is it better or worse for time stretching?

I mean, if we're just deliberately blinded with sample rates above 44.1 kHz (for better hardware sales and whatnot), then I won't use it further either. But I need to understand better first.

Post Reply
  • Information
  • Who is online

    Users browsing this forum: No registered users and 8 guests