Go Back   Cockos Incorporated Forums > REAPER Forums > REAPER Pre-Release Discussion

Reply
 
Thread Tools Display Modes
Old 01-09-2022, 01:42 PM   #1
flicker177
Human being with feelings
 
flicker177's Avatar
 
Join Date: Jan 2018
Location: Rochester, NY, USA
Posts: 97
Default FX oversampling and frequency response

Hi again,

I’ve been playing with the new, experimental FX oversampling a bit using Reaper 6.44+dev0108. I made a test project with one track and an instance of Reeq in the effects bin, enabled but set totally flat. I notice that oversampling this effect either individually or the complete FX chain affects the high frequency response; the more oversampling, the greater the high frequency rolloff.

This happens with any setting for the conversion from 16x sinc to R8-brain. At 2x, the effect is slight. At 8x the formerly flat frequency response is down 4dB at 20kHz, enough to somewhat dull the high end.

See the attached plot.

I wonder if this is a necessary artifact of the up/down sampling filters? It seems that it’s worth considering if you’re oversampling.

So...I did another test: The limiter I use is the Voxengo elephant which has selectable oversampling of its own. If I put that in my effects bin and set the level below its threshold, and don’t use Reaper’s FX oversampling, I can select 1x, 2x, 3x, and 4x oversampling in Elephant I don’t see this, I don’t know what might be different.

I’m concluding that oversampling may not always be a totally free lunch.

--Bill
Attached Images
File Type: jpg ostest-1x-8x.jpg (48.2 KB, 387 views)
__________________
--
Bill Thompson
https://BillThompson.us
flicker177 is offline   Reply With Quote
Old 01-09-2022, 01:56 PM   #2
Funkybot
Human being with feelings
 
Funkybot's Avatar
 
Join Date: Jul 2007
Location: New Joisey
Posts: 5,990
Default

Just an FYI - I think the devs prefer that items related to feature in dev get posted in the dev threads rather than making a new thread for each thing that comes up.

But yeah, it sounds like there may be some opportunities to use a steeper filter with a cutoff closer to nyquist. Maybe that can be implemented. And I wouldn't expect the filter cutoff itself to change to based on the OS level (I'd think the cutoff would be close to nyquist of current sample rate) but then I'm no DSP wiz so could be wrong on that.

That said, you're absolutely right in that there's no free lunch in audio. If there was, every plugin would sound amazing all the time. From what I've heard from developers, everything has trade-offs and finding the right ones is a delicate balance.
Funkybot is offline   Reply With Quote
Old 01-09-2022, 03:12 PM   #3
rafa1981
Human being with feelings
 
Join Date: Feb 2008
Posts: 188
Default

If that is the rolloff there is something misconfigured somewhere. r8brain is capable of much better quality.
rafa1981 is offline   Reply With Quote
Old 01-09-2022, 05:45 PM   #4
flicker177
Human being with feelings
 
flicker177's Avatar
 
Join Date: Jan 2018
Location: Rochester, NY, USA
Posts: 97
Default

Quote:
Originally Posted by rafa1981 View Post
If that is the rolloff there is something misconfigured somewhere. r8brain is capable of much better quality.
I don't think the rolloff is coming from R8brain because the result is exactly the same no matter what I choose for sample rate conversion.

--Bill
__________________
--
Bill Thompson
https://BillThompson.us
flicker177 is offline   Reply With Quote
Old 01-10-2022, 01:17 AM   #5
orbitfold
Human being with feelings
 
Join Date: Apr 2019
Posts: 136
Default

It isn't a necessary artefact of oversampling. Some roll off near Nyquist is necessary of course. But it should not depend on the multiple of the sampling rate you use for oversampling. So this is either a bug in oversampling or some problem with your procedure.
orbitfold is offline   Reply With Quote
Old 01-10-2022, 06:55 AM   #6
flicker177
Human being with feelings
 
flicker177's Avatar
 
Join Date: Jan 2018
Location: Rochester, NY, USA
Posts: 97
Default

Quote:
Originally Posted by orbitfold View Post
But it should not depend on the multiple of the sampling rate you use for oversampling. So this is either a bug in oversampling or some problem with your procedure.
I double checked this, I don't see a flaw in my procedure, but I simplified it.

Steps to reproduce:

1. Make a new empty project. Add a single track. Set Reaper's sample rate to 44.1k, bit depth to 24.

2. Put a sine wave generator in the track FX bin. Set to 1kHz, 0dB level. Set the FX bin oversampling to 8x. Start and stop playback to update the oversampling.

3. Note the level on the master buss is 0dB.

4. Change the sine wave frequency to 20kHz.

5. Note that the level on the master buss is -4.6dB

6. Set the FX bin oversampling to "No oversampling." Start and stop playback.

7. Note the level on the master buss returns to 0dB.

If you want to try this and don't have a sine wave generator, here's a simple one in JSFX:

Code:
desc: BT Sine Wave Generator
desc: Bills Sine wave Generator

slider1:-10<-120,0,1> Level (dB)
slider3:1000<20,20000,1> Frequency (Hz)

in_pin:none
out_pin:left output
out_pin:right output

@init

@slider
vol=2 ^ (slider1/6); 
freq=slider3;

@sample
// generate sine wave
adj = 2*$pi*freq/srate;
pos = pos+adj;
(pos >= 2.0*$pi) ? (pos -= 2.0*$pi); //wrap

tone = sin(pos)*vol; //sine wave output

spl0 = tone;
spl1 = tone;
I also made a 2-track project with sine wave generator feeding a track with a simple gain control and this track feeding the output. This was just to make sure that oversampling the sine wave generator was not an issue. The gain control has nothing in it that should affect frequency response. Oversampling the gain control effect using FX instance yields identical results.

Code:
desc:BT Gain
desc:simple gain control


slider1:0<-30,30,1>Gain (dB)

in_pin:left input
in_pin:right input
out_pin:left output
out_pin:right output

@init

@slider
  gain = 10^(slider1/20);

@sample
  spl0 *= gain;
  spl1 *= gain;
Certainly, oversampling this gain control should not affect frequency response. The only reason for the gain control is to have something to resample, you can't resample an empty FX bin.

The error is cumulative so if I resample the FX bin on both tracks, the rolloff at 20kHz is now -9.1dB.

If this response is not intended maybe this needs a look.

--Bill
__________________
--
Bill Thompson
https://BillThompson.us
flicker177 is offline   Reply With Quote
Old 01-10-2022, 04:46 PM   #7
Justin
Administrator
 
Justin's Avatar
 
Join Date: Jan 2005
Location: NYC
Posts: 15,721
Default

Quote:
Originally Posted by flicker177 View Post
5. Note that the level on the master buss is -4.6dB
The amount of passband attenuation here is dependent on the project resample mode. -4.6dB is what I'd expect if you're on 16pt sinc given a 20khz tone and 44khz samplerate. Try 192pt sinc and you'll find it's around -0.7dB. Try 384pt sinc or r8brain free and it should be negligible. (note you have to stop/start playback for that setting to take effect).
Justin is offline   Reply With Quote
Old 01-10-2022, 05:06 PM   #8
flicker177
Human being with feelings
 
flicker177's Avatar
 
Join Date: Jan 2018
Location: Rochester, NY, USA
Posts: 97
Default

Quote:
Originally Posted by Justin View Post
The amount of passband attenuation here is dependent on the project resample mode. -4.6dB is what I'd expect if you're on 16pt sinc given a 20khz tone and 44khz samplerate. Try 192pt sinc and you'll find it's around -0.7dB. Try 384pt sinc or r8brain free and it should be negligible. (note you have to stop/start playback for that setting to take effect).
I know that. I have set the sample rate conversion in the Render dialog to R8-brain, start and stop, and get the same result regardless of the sample rate conversion setting in the render dialog.

Do I need to change that setting somewhere else? I don't see where.

--Bill
__________________
--
Bill Thompson
https://BillThompson.us
flicker177 is offline   Reply With Quote
Old 01-10-2022, 05:08 PM   #9
Funkybot
Human being with feelings
 
Funkybot's Avatar
 
Join Date: Jul 2007
Location: New Joisey
Posts: 5,990
Default

Quote:
Originally Posted by flicker177 View Post
I know that. I have set the sample rate conversion in the Render dialog to R8-brain, start and stop, and get the same result regardless of the sample rate conversion setting in the render dialog.

Do I need to change that setting somewhere else? I don't see where.

--Bill
Maybe post the test project here (dropbox or similar link should suffice).
Funkybot is offline   Reply With Quote
Old 01-10-2022, 06:20 PM   #10
Senior Negative
Human being with feelings
 
Join Date: Jan 2022
Posts: 6
Default

Render is wrong place if i understand that is place you changing. Change in file - project settings. In render you only change what is used in rendering whole project. Your sine wave example would be not affected
Senior Negative is offline   Reply With Quote
Old 01-10-2022, 06:57 PM   #11
flicker177
Human being with feelings
 
flicker177's Avatar
 
Join Date: Jan 2018
Location: Rochester, NY, USA
Posts: 97
Default

Quote:
Originally Posted by Senior Negative View Post
Render is wrong place if i understand that is place you changing. Change in file - project settings. In render you only change what is used in rendering whole project. Your sine wave example would be not affected
OK, I checked the setting in File|Project Settings|Playback Resample Mode. That was set to 64 bit sinc. I'd never looked at that setting. It looks like 64 bit sinc gives me an even worse rolloff than 16 bit sinc. If I set it to R8-Brain the rolloff is gone and the response is flat.

Sorry for the false alarm although before this feature is officially released it might be a good idea to have a better default or some alert to change it for flat response to avoid confusion to easily confused people like me ;-)

I don't see any reason to use anything other than R8-Brain and that this would be a great default to avoid issues.

Thanks much to all!

--Bill
__________________
--
Bill Thompson
https://BillThompson.us

Last edited by flicker177; 01-10-2022 at 07:05 PM.
flicker177 is offline   Reply With Quote
Old 01-11-2022, 03:37 AM   #12
orbitfold
Human being with feelings
 
Join Date: Apr 2019
Posts: 136
Default

That still doesn't explain anything if I understood your initial problem correctly. It would seem that the downsampling filter is applied multiple times depending on the number of times you oversample. Which shouldn't be the case. Maybe I'm misunderstanding something.
orbitfold is offline   Reply With Quote
Old 01-11-2022, 07:31 AM   #13
flicker177
Human being with feelings
 
flicker177's Avatar
 
Join Date: Jan 2018
Location: Rochester, NY, USA
Posts: 97
Default

Quote:
Originally Posted by orbitfold View Post
That still doesn't explain anything if I understood your initial problem correctly. It would seem that the downsampling filter is applied multiple times depending on the number of times you oversample. Which shouldn't be the case. Maybe I'm misunderstanding something.
I'm sure I'm misunderstanding a lot ;-)

There are others here who understand oversampling and its filters better than I and perhaps someone can chime in to explain it. In the mean time, with the playback resample mode set to R8-Brain the oversampling seems to work very well and it uses less CPU than 64 bit sinc! I no longer see any use for the other resample algorithms. (but I may be missing something there too)

--Bill
__________________
--
Bill Thompson
https://BillThompson.us
flicker177 is offline   Reply With Quote
Old 01-11-2022, 01:15 PM   #14
Justin
Administrator
 
Justin's Avatar
 
Join Date: Jan 2005
Location: NYC
Posts: 15,721
Default

Quote:
Originally Posted by flicker177 View Post
I'm sure I'm misunderstanding a lot ;-)

There are others here who understand oversampling and its filters better than I and perhaps someone can chime in to explain it. In the mean time, with the playback resample mode set to R8-Brain the oversampling seems to work very well and it uses less CPU than 64 bit sinc! I no longer see any use for the other resample algorithms. (but I may be missing something there too)

--Bill
The other algorithms have lower latency, fwiw
Justin is offline   Reply With Quote
Old 01-11-2022, 06:46 PM   #15
flicker177
Human being with feelings
 
flicker177's Avatar
 
Join Date: Jan 2018
Location: Rochester, NY, USA
Posts: 97
Default

Quote:
Originally Posted by Justin View Post
The other algorithms have lower latency, fwiw
Interesting, I hadn't even looked at that. If you're trying to do some kind of real-time monitoring it suddenly gets *very* important.

It's really interesting that you can provide good oversampling to effects that don't provide it, especially non-linear stuff that generates aliasing which I've run into many times.

Thanks, Justin for all the hard work!

--Bill
__________________
--
Bill Thompson
https://BillThompson.us
flicker177 is offline   Reply With Quote
Old 01-12-2022, 10:26 AM   #16
daxliniere
Human being with feelings
 
daxliniere's Avatar
 
Join Date: Nov 2008
Location: London, UK
Posts: 2,581
Default

Quote:
Originally Posted by Justin View Post
The other algorithms have lower latency, fwiw
Could you report the latency in the project settings next to the descriptions (Medium quality, slow, very slow, highest quality), please?
__________________
Puzzle Factory Sound Studios, London [Website] [Instagram]
[AMD 5800X, 32Gb RAM, Win10x64, NVidia GTX1080ti, UAD2-OCTO, FireFaceUCX, REAPER x64]
[Feature request: More details in Undo History]
daxliniere is offline   Reply With Quote
Old 01-13-2022, 02:59 PM   #17
Justin
Administrator
 
Justin's Avatar
 
Join Date: Jan 2005
Location: NYC
Posts: 15,721
Default

Quote:
Originally Posted by daxliniere View Post
Could you report the latency in the project settings next to the descriptions (Medium quality, slow, very slow, highest quality), please?
Hmm perhaps somewhere. The main thing is that they increase along with perceived quality (and r8brain trades speed for even more latency)
Justin is offline   Reply With Quote
Old 01-13-2022, 03:36 PM   #18
daxliniere
Human being with feelings
 
daxliniere's Avatar
 
Join Date: Nov 2008
Location: London, UK
Posts: 2,581
Default

Cool. So presumably if you use have a fast enough computer, you could use the higher sinc modes without additional latency.

Interestingly, I really liked what 768pt sinc (@4x OS) added to a master I was working on. r8brain sounded more like the non-oversampled master. Fractionally better, but same 'flavour'.
__________________
Puzzle Factory Sound Studios, London [Website] [Instagram]
[AMD 5800X, 32Gb RAM, Win10x64, NVidia GTX1080ti, UAD2-OCTO, FireFaceUCX, REAPER x64]
[Feature request: More details in Undo History]
daxliniere is offline   Reply With Quote
Old 01-13-2022, 04:22 PM   #19
Justin
Administrator
 
Justin's Avatar
 
Join Date: Jan 2005
Location: NYC
Posts: 15,721
Default

Quote:
Originally Posted by daxliniere View Post
Cool. So presumably if you use have a fast enough computer, you could use the higher sinc modes without additional latency.
They still add latency. They don't necessarily have to add much, but if you want to be linear phase you do have to add a decent amount.
Justin is offline   Reply With Quote
Old 01-13-2022, 04:29 PM   #20
daxliniere
Human being with feelings
 
daxliniere's Avatar
 
Join Date: Nov 2008
Location: London, UK
Posts: 2,581
Default

Hmmm... What does minimum phase resampling sound like?
__________________
Puzzle Factory Sound Studios, London [Website] [Instagram]
[AMD 5800X, 32Gb RAM, Win10x64, NVidia GTX1080ti, UAD2-OCTO, FireFaceUCX, REAPER x64]
[Feature request: More details in Undo History]
daxliniere is offline   Reply With Quote
Old 01-14-2022, 12:39 AM   #21
rafa1981
Human being with feelings
 
Join Date: Feb 2008
Posts: 188
Default

r8brain has a minimum-phase mode if I remember correctly.
rafa1981 is offline   Reply With Quote
Old 01-14-2022, 08:03 AM   #22
Tale
Human being with feelings
 
Tale's Avatar
 
Join Date: Jul 2008
Location: The Netherlands
Posts: 3,645
Default

AFAICT r8brain as used in REAPER is linear phase, because its impulse response looks symmetrical.
Tale is offline   Reply With Quote
Old 01-14-2022, 10:33 AM   #23
Justin
Administrator
 
Justin's Avatar
 
Join Date: Jan 2005
Location: NYC
Posts: 15,721
Default

We could enable r8brain's minimum phase mode, but meh (it also has some caveats and doesn't decrease the latency by much)

Last edited by Justin; 01-14-2022 at 10:55 AM.
Justin is offline   Reply With Quote
Old 01-17-2022, 10:19 AM   #24
Arthur McArthur
Human being with feelings
 
Arthur McArthur's Avatar
 
Join Date: Sep 2016
Location: Toronto
Posts: 744
Default

r8brain minimum phase would be a welcome addition.

Minimum phase oversampling can sometimes beat out linear phase on transient heavy sources like drums.
Arthur McArthur is offline   Reply With Quote
Old 01-17-2022, 01:43 PM   #25
Lynx_TWO
Human being with feelings
 
Lynx_TWO's Avatar
 
Join Date: Dec 2019
Location: St Petersburg FL
Posts: 996
Default

Dumb Question: How do you enable FX oversampling besides changing the project rate? Using the latest 4.6 RC with the option, but can't seem to find it anywhere.
__________________
My mixes from the Cambridge multitracks library
SoundCloud link & Youtube (ThemTube?) link
My preferred adjectives are “Handsome” and “Brilliant”
Lynx_TWO is offline   Reply With Quote
Old 01-17-2022, 03:57 PM   #26
nofish
Human being with feelings
 
nofish's Avatar
 
Join Date: Oct 2007
Location: home is where the heart is
Posts: 12,096
Default

Quote:
Originally Posted by Lynx_TWO View Post
Dumb Question: How do you enable FX oversampling besides changing the project rate? Using the latest 4.6 RC with the option, but can't seem to find it anywhere.
Looks like FX oversampling didn't make it in the v6.46rc (from a quick peek at the changelog).
nofish is offline   Reply With Quote
Old 01-17-2022, 08:00 PM   #27
Lynx_TWO
Human being with feelings
 
Lynx_TWO's Avatar
 
Join Date: Dec 2019
Location: St Petersburg FL
Posts: 996
Default

Quote:
Originally Posted by nofish View Post
Looks like FX oversampling didn't make it in the v6.46rc (from a quick peek at the changelog).
Ahh. Thanks!
__________________
My mixes from the Cambridge multitracks library
SoundCloud link & Youtube (ThemTube?) link
My preferred adjectives are “Handsome” and “Brilliant”
Lynx_TWO is offline   Reply With Quote
Old 01-22-2022, 08:42 AM   #28
beingmf
Human being with feelings
 
beingmf's Avatar
 
Join Date: Jul 2007
Location: Jazz City
Posts: 5,065
Default

Quote:
Originally Posted by nofish View Post
Looks like FX oversampling didn't make it in the v6.46rc (from a quick peek at the changelog).
I wonder why. It works like a charm here!?
__________________
Windows 10x64 | AMD Ryzen 3700X | ATI FirePro 2100 | Marian Seraph AD2, 4.3.8 | Yamaha Steinberg MR816x
"If I can hear well, then everything I do is right" (Allen Sides)
beingmf is offline   Reply With Quote
Old 01-22-2022, 10:47 AM   #29
Lynx_TWO
Human being with feelings
 
Lynx_TWO's Avatar
 
Join Date: Dec 2019
Location: St Petersburg FL
Posts: 996
Default

Quote:
Originally Posted by beingmf View Post
I wonder why. It works like a charm here!?
Maybe because of the filtering issue people were running into; 20k was down like 4db when enabled
__________________
My mixes from the Cambridge multitracks library
SoundCloud link & Youtube (ThemTube?) link
My preferred adjectives are “Handsome” and “Brilliant”
Lynx_TWO is offline   Reply With Quote
Old 01-27-2022, 03:19 AM   #30
inoto
Human being with feelings
 
Join Date: Nov 2021
Posts: 47
Default

Quote:
Originally Posted by Justin View Post
The other algorithms have lower latency, fwiw
Does this apply also to regular resampling during playback?
Or are the latencies equalized internally somehow?

Because I was curious and did a test:
1. set soundcard ASIO to 44.1kHz in Reaper
2. 96kHz kick drum sample on track 1
3. switch between playback resampling modes and record the kick with loopback

There was no difference in latency between the 'Sinc' modes and 'r8brain'. Exact same timing down to the sample.
The 'Point' and 'Linear' modes had maybe +- 1 sample differences.

This was with v6.46 stable, BTW.

Last edited by inoto; 01-27-2022 at 03:30 AM.
inoto is offline   Reply With Quote
Old 01-28-2022, 08:46 PM   #31
Lynx_TWO
Human being with feelings
 
Lynx_TWO's Avatar
 
Join Date: Dec 2019
Location: St Petersburg FL
Posts: 996
Default

I believe it does. Probably best to use r8brain for playback AND rendering if you can
__________________
My mixes from the Cambridge multitracks library
SoundCloud link & Youtube (ThemTube?) link
My preferred adjectives are “Handsome” and “Brilliant”
Lynx_TWO is offline   Reply With Quote
Old 01-29-2022, 11:38 AM   #32
flicker177
Human being with feelings
 
flicker177's Avatar
 
Join Date: Jan 2018
Location: Rochester, NY, USA
Posts: 97
Default

Quote:
Originally Posted by inoto View Post
There was no difference in latency between the 'Sinc' modes and 'r8brain'. Exact same timing down to the sample.
I think you're not seeing the latency because its compensated for. It doesn't seem affected by the oversampling factor but the different modes have very different latency.

Using the performance monitor (<ctrl><alt>P) to see the compensation shows this:

Code:
Mode      Delay (samples)
r8brain   3302
sinc768   434
sinc384   218
sinc64    38
--Bill
__________________
--
Bill Thompson
https://BillThompson.us
flicker177 is offline   Reply With Quote
Old 01-29-2022, 05:20 PM   #33
daxliniere
Human being with feelings
 
daxliniere's Avatar
 
Join Date: Nov 2008
Location: London, UK
Posts: 2,581
Default

Quote:
Originally Posted by Lynx_TWO View Post
Maybe because of the filtering issue people were running into; 20k was down like 4db when enabled
That's not an 'issue people were running into', it was that one user had their resample quality set very low (16pt sinc) and the HF roll-off was to be expected.


Quote:
Originally Posted by flicker177 View Post
Code:
Mode      Delay (samples)
r8brain   3302
sinc768   434
sinc384   218
sinc64    38
Awesome! Thanks Bill!
(BTW, I was in Rochester last Xmas/NYE with family)
__________________
Puzzle Factory Sound Studios, London [Website] [Instagram]
[AMD 5800X, 32Gb RAM, Win10x64, NVidia GTX1080ti, UAD2-OCTO, FireFaceUCX, REAPER x64]
[Feature request: More details in Undo History]
daxliniere is offline   Reply With Quote
Old 01-30-2022, 07:38 AM   #34
Lynx_TWO
Human being with feelings
 
Lynx_TWO's Avatar
 
Join Date: Dec 2019
Location: St Petersburg FL
Posts: 996
Default

Quote:
Originally Posted by daxliniere View Post
That's not an 'issue people were running into', it was that one user had their resample quality set very low (16pt sinc) and the HF roll-off was to be expected.
Ahhh gotchya! Yes I have mine set on r8brain for playback and rendering so I’ve noticed no issues. I use a mixing template with a channel strip plugin on each track (with the noise turned all the way down, just using it for that little extra harmonic distortion and tiny differences between channels) and so far it “feels” like adding 2x oversampling to each instance on each track adds a tiny bit of clarity without slowing down rendering significantly. Could totally be the placebo effect though; I’d have to do a zero sum to know for sure
__________________
My mixes from the Cambridge multitracks library
SoundCloud link & Youtube (ThemTube?) link
My preferred adjectives are “Handsome” and “Brilliant”
Lynx_TWO is offline   Reply With Quote
Old 01-30-2022, 05:30 PM   #35
daxliniere
Human being with feelings
 
daxliniere's Avatar
 
Join Date: Nov 2008
Location: London, UK
Posts: 2,581
Default

Hey Lynx_Two,
Yes, definitely try it out with some A/B testing. I was surprised how good a recent master I'd done sounded with 4x OS using 768pt sinc. r8brain sounded close most similar to the original, non-oversampled master, but slightly 'better'. Sinc seemed to add something really nice, I liked it a lot. YMMV, but definitely give it a shot.

Have fun!
Dax.
__________________
Puzzle Factory Sound Studios, London [Website] [Instagram]
[AMD 5800X, 32Gb RAM, Win10x64, NVidia GTX1080ti, UAD2-OCTO, FireFaceUCX, REAPER x64]
[Feature request: More details in Undo History]
daxliniere is offline   Reply With Quote
Old 07-08-2022, 04:23 AM   #36
vsgrt
Human being with feelings
 
Join Date: May 2017
Posts: 186
Default

Hmm, does r8brain really add 3302 samples of latency?
That seems like a lot.
It says "highest quality, fast" - is that fast as in CPU terms, and not latency?

So if i have any non-native samplerate samples in my project, i get 3302 latency of SRC on playback?

Also, i don't see any PDC changes when oversampling with Reapers native OS - i assume the latency is "hidden", and if so, what is the exact latency for a 2x OS?
vsgrt is offline   Reply With Quote
Old 07-10-2022, 06:42 AM   #37
Justin
Administrator
 
Justin's Avatar
 
Join Date: Jan 2005
Location: NYC
Posts: 15,721
Default

Quote:
Originally Posted by vsgrt View Post
Hmm, does r8brain really add 3302 samples of latency?
That seems like a lot.
It says "highest quality, fast" - is that fast as in CPU terms, and not latency?

So if i have any non-native samplerate samples in my project, i get 3302 latency of SRC on playback?

Also, i don't see any PDC changes when oversampling with Reapers native OS - i assume the latency is "hidden", and if so, what is the exact latency for a 2x OS?
r8brain has the most latency of all of the SRC modes. For media resampling it doesn’t affect much (look ahead on media is easy!), but for FX oversampling it is more consequential. The sinc modes use more CPU but have lower latency (ranging from 30-500 samples or so, the exact values you can test)
Justin is offline   Reply With Quote
Old 07-12-2022, 11:57 PM   #38
Lynx_TWO
Human being with feelings
 
Lynx_TWO's Avatar
 
Join Date: Dec 2019
Location: St Petersburg FL
Posts: 996
Default

Quote:
Originally Posted by vsgrt View Post
Hmm, does r8brain really add 3302 samples of latency?
That seems like a lot.
It says "highest quality, fast" - is that fast as in CPU terms, and not latency?

So if i have any non-native samplerate samples in my project, i get 3302 latency of SRC on playback?

Also, i don't see any PDC changes when oversampling with Reapers native OS - i assume the latency is "hidden", and if so, what is the exact latency for a 2x OS?
Yes, Reaper automatically calculates for the higher latency. It’s probably not the best to use if you’re recording punch-in, punch-outs, but for mixing and mastering purposes it’s the best way to go currently. Also, at 48khz, 3302 samples is less than a 68ms delay. For higher sample rates it’d be even less.
__________________
My mixes from the Cambridge multitracks library
SoundCloud link & Youtube (ThemTube?) link
My preferred adjectives are “Handsome” and “Brilliant”
Lynx_TWO is offline   Reply With Quote
Old 07-14-2022, 04:28 AM   #39
DuX
Human being with feelings
 
DuX's Avatar
 
Join Date: Jan 2007
Location: Underworld
Posts: 1,188
Default

Quote:
Originally Posted by flicker177 View Post

Thanks, Justin for all the hard work!

--Bill
And also big thanks to Alexey Vaneev - Voxengo for open sourcing his incredibly good SRC algorithm.
__________________
Goodnight, thank you, and may your God go with you.
DuX is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 12:55 PM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.