4.0alpha7a - Resampling causes crackling :(

[Dstruct]

Load this project: http://forum.cockos.com/attachment.p...8&d=1455568380


-> set your soundcard samplerate to 44.1 kHz

-> playback (so we have a 48kHz -> 44.1kHz conversion going on)


=> Crackles/glitches, no matter which resample mode


Crackles are also there in offline render.


FIXED (5.20pre4)

[Subz]

EDIT::


I also am getting a light crackle at 44.1 & none at 48 with the linked project

Subz

[Dstruct]

Yup, not sure if it's new to 4.alpha as I don't have 3.x installed here to test ...

[arkima]

Hey,

Yeah crackles (/tiny pops?) audible at my end from the supplied project & media. However still audible @44100 and 48000.
Copy-paste then glue rendered - still audible (don't know if that should make a difference).

-a


oops:

Reaper (x64) 4.0alpha7a
XP 64
Line6 UX8 (usb)

[Dstruct]

Originally Posted by arkima

Hey,

Yeah crackles (/tiny pops?) audible at my end from the supplied project & media. However still audible @44100 and 48000.

It (the project with the sine loop) should loop fine at 48 kHz session. Only when running it at 44.1 kHz it causes these glitches here.


FIXED (5.20pre4)

[arkima]

Originally Posted by Dstruct

It (the project with the sine loop) should loop fine at 48 kHz session. Only when running it at 44.1 kHz it causes these glitches here.

Hi Dstruct,

Here's the thing - Yeah, I just tested it out on the (x86)v3.73 and sounds clear (I really cranked it too - no crackle that I could hear) @48kHz but there as you mention @44.1kHz (so it's resampling) - I can hear it.

But no matter what at my end I can still hear the crackle testing it out on 3.73 (x64), 4.0 alpha7a (x86) and 4.0alpha7a (x64).

?

I did testing using headphones. And not 100% certain but it did seem worse playing back @44.1kHz in general - here anyway..

Hope that's some kind of use..

-a

[Dstruct]

I don't really get what you want to say.


You mean you get crackles in "3.73 (x64), 4.0 alpha7a (x86) and 4.0alpha7a (x64)" at 48 kHz session too?

[Subz]

3.74Pre4 & 3.52 act the same as 4Alpha7a for me

all 3 crackle at 44.1 but play fine at 48

Subz

(Win7 x86)

[musicbynumbers]

I really doubt this is it but perhaps it's to do with the new sample rate snapping stuff (behind the scenes that is) because I know that it's not because the items have been moved but maybe something about how the samples are being "read back" in this mode.

Still, this is doubtful but worth a test with project sample rate off and the new sample snap box un ticked..

[Dstruct]

It's not caused by the resampling algorithm. It's the item looping that's buggy.


Maybe the bug was introduced in 3.3 where this http://forum.cockos.com/project.php?issueid=1939 got fixed ...

[Subz]

this needs to go in the tracker & the V3 bug forum!! (IMHO)

Subz

[Dstruct]

Put into bug tracker: http://forum.cockos.com/project.php?issueid=3125 (please vote/confirm) ...

FIXED (5.20pre4)

[arkima]

Originally Posted by Dstruct

I don't really get what you want to say.


You mean you get crackles in "3.73 (x64), 4.0 alpha7a (x86) and 4.0alpha7a (x64)" at 48 kHz session too?

My apologies Dstruct, that's what I thought I heard at that time.(& embarrassed smiley goes here)

So this time taking a breath and carefully running through each and re-checking:

-All are fine @48 kHz playback and at loop. (as expected)
-All @44.1 kHz contain small crackle/pop through play back and a pop/click at the loop point is noticeable.

Sorry - was just trying to help.

[Dstruct]

Alright, thanks. Please confirm in the tracker too ...

[arkima]

Done.

[mirceablue]

yes I can confirm that that I have crackles too since the first version of 4 alpha, RME 9652 card here at 44.1 khz.

[Subz]

Originally Posted by arkima

Done.

FYI there are [yes] / [no] buttons on tracker posts thats how you confirm

posting in tracker threads with no relevant info may result in deletion of said post



Subz

[Dstruct]

Originally Posted by mirceablue

yes I can confirm that that I have crackles too since the first version of 4 alpha, RME 9652 card here at 44.1 khz.

But this issue isn't new to 4.alpha!

[Dstruct]

Originally Posted by arkima

Done.

Not really. You need to press Yes/No to vote. No reply with text needed.

[arkima]

Originally Posted by Dstruct

Not really. You need to press Yes/No to vote. No reply with text needed.

Woops, very shabby of me. Voted/confirmed..

Thanks for pulling me up Dstruct.

-a

[Ollie]

- I'm not getting any crackling at 44.1 or 48kHz, also not at the loop point.

- The output render WAV file you provided is clean (no audible artifacts or visible glitches on the waveform display, hence no rendered artifacts)

So there must be more to it than playing back the test project @44.1kHz, and it looks a bit like the problem could be rather on the playback side outside of REAPER than within. OS is ruled out (I used Vista 32 and you guys are using all other Windowses) - it looks like at least 2 of you guys are using an RME card, what else do you have in common?

[Subz]

Originally Posted by Ollie

- I'm not getting any crackling at 44.1 or 48kHz, also not at the loop point.

- The output render WAV file you provided is clean (no audible artifacts or visible glitches on the waveform display, hence no rendered artifacts)

So there must be more to it than playing back the test project @44.1kHz, and it looks a bit like the problem could be rather on the playback side outside of REAPER than within. OS is ruled out (I used Vista 32 and you guys are using all other Windowses) - it looks like at least 2 of you guys are using an RME card, what else do you have in common?

the "audible artifacts" are very quite! you need to turn the audio up a bit to notice it,

there is a quite obvious one in the wav file at 4 seconds in

i think they are sub sample errors so you need to zoom in close to find it

this is in the wav

[IMG]http://img225.**************/img225/7501/44112760.png[/IMG]

Subz

[Dstruct]

Originally Posted by Ollie

- I'm not getting any crackling at 44.1 or 48kHz, also not at the loop point.

- The output render WAV file you provided is clean (no audible artifacts or visible glitches on the waveform display, hence no rendered artifacts)

Then you really need to see a doctor to check your ears!

[Subz]

Originally Posted by Dstruct

Then you really need to see a doctor to check your ears!

i wouldn't go that far!

first time i played it i did not notice it,

& i guess with Ollie being so mad busy dealing with this mad house of a forum he only quickly tested with low volume

or he just has a naturally crackly setup at home

Subz

[Ollie]

Originally Posted by Subz

the "audible artifacts" are very quite! you need to turn the audio up a bit to notice it,

there is a quite obvious one in the wav file at 4 seconds in

i think they are sub sample errors so you need to zoom in close to find it

this is in the wav

[IMG]http://img225.**************/img225/7501/44112760.png[/IMG]

Subz

Look how helpful a screenshot and/or some detaled information can be!

[Dstruct]

Still there in 5.16pre1.

See attached project (fine at 48kHz, crackles at 44.1kHz)

FIXED (5.20pre4)

[Justin]

Working on this now! Hopefully the fix won't break the world.

[Dstruct]

Ok, thanks!

[ELP]

issue seems to fixed with 5.20pre4,

at least, I can not hear crackles anymore, like before.

[Dstruct]

Yes, crackles are fixed. But I think I'm still able to hear some overtones/distortion that doesn't sound right to me!? I think it should sound cleaner. But might be the speakers too.


Also see this new attached project:

track 1: item looped normally
track 2: item looped using "loop section of item source"


Results:

at 48kHz: fine (nulls out to -inf)

at 44.1kHz: only the first pass (when starting playback at about the middle of the items) nulls out to -inf. all following passes only to -44 dBFS. shouldn't this be better/lower? why is the first (half) pass nulling out to -inf?


Just hardcore testing

[Justin]

Originally Posted by Dstruct

Yes, crackles are fixed. But I think I'm still able to hear some overtones/distortion that doesn't sound right to me!? I think it should sound cleaner. But might be the speakers too.

What audio output mode are you using? What hardware?

(I mention this because anything going through the Win7 kernel mixer (WaveOut, DSound, etc) does sound distorted at 44khz if the audio device is natively 48khz).

Also see this new attached project:

track 1: item looped normally
track 2: item looped using "loop section of item source"


Results:

at 48kHz: fine (nulls out to -inf)

at 44.1kHz: only the first pass (when starting playback at about the middle of the items) nulls out to -inf. all following passes only to -44 dBFS. shouldn't this be better/lower? why is the first (half) pass nulling out to -inf?


Just hardcore testing

I didn't look at this project, but I did a test where I glued the original project at 48khz, then switched to 44khz and compared it to the looped version. There are differences after the first loop, and this is to be expected, due to the various compromises made by samplerate conversion. Hopefully it's not glitchy, though anyway...

[Dstruct]

Originally Posted by Justin

What audio output mode are you using? What hardware?

(I mention this because anything going through the Win7 kernel mixer (WaveOut, DSound, etc) does sound distorted at 44khz if the audio device is natively 48khz).

ASIO, RME HDSP9632, Yamaha MSP7, Beyerdynamic dt-990 headphones

There is some sort of buzz but it's already there in the original 48kHz sample (if you bosst >500Hz). I took this as clean sine from Vember Audio Surge Vsti.

Yeah, it seems to be impossible in REAPER to cut a single cycle sine wave from an audio file. If you try to do this you'll get this noize/buzz above 500Hz!? Can you get a clean one?

Snap to samples is enabled + force project tempo ... to occur on whole samples.

ReaSynth is the same. Can't cut a single cycle from it without buzz on loop.

Originally Posted by Justin

I didn't look at this project, but I did a test where I glued the original project at 48khz, then switched to 44khz and compared it to the looped version. There are differences after the first loop, and this is to be expected, due to the various compromises made by samplerate conversion. Hopefully it's not glitchy, though anyway...

No, not glitchy. Won't do this in real usage anyway.

[Justin]

Originally Posted by Dstruct

If you try to do this you'll get this noize/buzz above 500Hz!? Can you get a clean one?

Snap to samples is enabled + force project tempo ... to occur on whole samples.

ReaSynth is the same. Can't cut a single cycle from it without buzz on loop.

I think you need to have the wavelength be an exact multiple of samples.

So for example, at 48000 Hz, a 440 Hz sine wave would be 109.0909090909091 samples long = would glitch, but a 480 Hz sine wave would be exactly 100 samples long and should be OK.

Here's a 480 Hz sinewave I made by gluing the output of the tone generator, it looks clean looped at 48kHz on a spectrum analyzer:

http://1014.org/shiz/shup/01-160216_...1-glued-01.wav

[Dstruct]

Ok, solved then! This 480 Hz sine is looping fine! Thanks!

[Justin]

I'm going to do some research to see if I can make looped items behave better with resampling too. It might be worthwhile...

[Dstruct]

Ok, cool