Go Back   Cockos Incorporated Forums > REAPER Forums > Recording Technologies and Techniques

Reply
 
Thread Tools Display Modes
Old 03-30-2014, 11:35 AM   #1
frets
Human being with feelings
 
Join Date: Feb 2014
Location: canada
Posts: 7
Default resampling a 48khz song to 44.1 - stems vs entire mix

I have a song with many tracks in reaper.
There is a mixture of 44.1khz and 48khz tracks within the song.
I eventually want to release at 44.1khz.

What is better:

A: just render the song in reaper at 44.1khz.
[ie. the resampling will be done on each track individually and then rendered together]

B: Render the song at 48khz. Then open that mixdown in a new reaper project, and render to 44.1khz.
[ie. run the resampling algorithm once on the full mix]

Why?
frets is offline   Reply With Quote
Old 03-30-2014, 10:14 PM   #2
yep
Human being with feelings
 
Join Date: Aug 2006
Posts: 2,019
Default

Quote:
Originally Posted by frets View Post
I have a song with many tracks in reaper.
There is a mixture of 44.1khz and 48khz tracks within the song.
I eventually want to release at 44.1khz.

What is better:

A: just render the song in reaper at 44.1khz.
[ie. the resampling will be done on each track individually and then rendered together]

B: Render the song at 48khz. Then open that mixdown in a new reaper project, and render to 44.1khz.
[ie. run the resampling algorithm once on the full mix]

Why?
Do all your mixing and processing at the "source" sample rate (48, in this case), then make the final step be a render to the "target" sample rate and bit depth.

The "why" is pretty simple: every conversion step introduces at least some theoretical error/distortion, so you are better off having the fewest possible sample-rate conversions. The ideal number of times to downsample is zero. The next-best is once. So if you have a choice between downsampling a bunch of times, versus downsampling once, do it once, and last. You'll probably never hear a difference either way, but it's one less thing to worry about.
yep is offline   Reply With Quote
Old 03-31-2014, 08:27 AM   #3
audioguy_on_ca
Human being with feelings
 
Join Date: Apr 2008
Posts: 259
Default This should be Stickied

Mods?
Yep, yep nailed it. so should y'all, powers that be.
__________________
"We're like geeky and shit: I'm hanging out on a message board about a recording program. It's kinda expected to be all up in arms about something as ridiculously abstract as 24->16 bit dithering..." -RPR usr nickm
audioguy_on_ca is offline   Reply With Quote
Old 03-31-2014, 09:12 AM   #4
serr
Human being with feelings
 
Join Date: Sep 2010
Posts: 12,562
Default

Quote:
Originally Posted by frets View Post
I have a song with many tracks in reaper.
There is a mixture of 44.1khz and 48khz tracks within the song.
I eventually want to release at 44.1khz.

What is better:

A: just render the song in reaper at 44.1khz.
[ie. the resampling will be done on each track individually and then rendered together]

B: Render the song at 48khz. Then open that mixdown in a new reaper project, and render to 44.1khz.
[ie. run the resampling algorithm once on the full mix]

Why?
Since you have a mix of 48k & 44.1k source material, aiming for your master in 44.1k is the most reasonable.

***Note that if you actually have MOST (and/or the most important) of your tracks at 48k with just a few at 44.1k, then my advice would be the opposite and I would advise you to release at 24 bit 48k instead. Also, I'm assuming this is audio only. If this were a soundtrack for video, I would advise you to release at 24 bit 48k instead as it's the default choice for video. ***

Some points:
- 44.1k is a more accessible format (even if considered 'portable quality' nowadays)
- The sample rate conversion between the lower resolution rates (44.1k & 48k) is the most difficult/damaging
- Upsampling both sources to 96k, while slightly better preserving the 48k source would end up using 4 times the space for almost no advantage (even to the snobbiest audiophile - the source would need to be 24/96 to take advantage of the high def release format)

So...
A. 48k source tracks slightly degraded to 44.1k but then everything else untouched and more important, the final master not subjected to sample rate conversion.

vs.

B. 44.1k source SRC to 48k, mix and master at 48k but then degrade the entire master with the most difficult conversion: 48k to 44.1k

Hands down go with choice A!

That you are using 24 bit (regardless of the sample rate) is actually the much bigger factor. (And preserving that dynamic range with Reaper's 64 bit fp mix engine.)

It goes without saying to keep everything at 24 bit (and thus make your final release 24 bit 44.1k). Then follow up and make a "portable master" at 16 bit 44.1k.

Note: for best results, render/freeze your 48k tracks to 44.1k ahead of time with your best conversion algorithms vs. letting Reaper convert them on the fly.

Last edited by serr; 03-31-2014 at 11:38 AM.
serr is offline   Reply With Quote
Old 03-31-2014, 01:55 PM   #5
frets
Human being with feelings
 
Join Date: Feb 2014
Location: canada
Posts: 7
Default

Ah yep, it seems so simple once you explain it in terms of doing the fewest sample conversions possible.

For anyone following along at home, here`s how I implemented it in Reaper:

The "Project Sample Rate" is 48000 Hz and the "Render Resample Mode" is set to EXTREME.
[File -> Project Settings -> Project Settings Tab]

Upon rendering, I chose:
Sample Rate: [44100 Hz].
Use project sample rate for mixing and FX/synth processing: [checked]
Resample mode (if needed): [Extreme].
frets is offline   Reply With Quote
Old 03-31-2014, 01:58 PM   #6
frets
Human being with feelings
 
Join Date: Feb 2014
Location: canada
Posts: 7
Default

Quote:
Originally Posted by serr View Post
... and more important, the final master not subjected to sample rate conversion.
Why is it more important for this to not happen on the final master? This is essentially the opposite of yep's recommendation to reduce the resample count to one by doing it once at the end.

For the record, though, I could not actually hear the difference between any of it. I couldn't even pretend to fool myself into thinking I heard it
frets is offline   Reply With Quote
Old 03-31-2014, 03:39 PM   #7
serr
Human being with feelings
 
Join Date: Sep 2010
Posts: 12,562
Default

Quote:
Originally Posted by frets View Post
Why is it more important for this to not happen on the final master? This is essentially the opposite of yep's recommendation to reduce the resample count to one by doing it once at the end.

For the record, though, I could not actually hear the difference between any of it. I couldn't even pretend to fool myself into thinking I heard it
The final mix is a dense thing with nuance at many levels.

A single raw track that is to be a component of a mix is a lot more forgiving. You can tweak or 'repair' that and not hear anything in the final mix.

Tweak or 'repair' the final mix though... It becomes delicate.


It's really all about recording at the highest quality of your equipment and keeping everything as clean and simple as possible. The whole point (for me anyway) is to be able to 'set it and forget it' and just record and mix without going back and second guessing if something was set right and if there's now some tiny difference I need to try and listen for.

Just to be fair (and this is what yep would have been going for), you could try converting everything to 96k and mixing and mastering at that. (Then make a 44.1k portable master from that.)

I'm avoiding the lower res conversions with lower res source material. Converting to higher res though can give you working room and retain the original fidelity.

I've done some tests before with conversions. Do 10 or 100 in a row back and forth to really highlight if something degrades due to the process. Back and forth with 44.1k & 48k degrades. Upsampling and going between 88.2k and 96k does almost nothing funny. After the very 1st conversion to 96k from 192k, every back and forth between 192k and 96k after that can be directly phase cancelled to zero after 100 conversions (ie they are digitally identical).

Last edited by serr; 03-31-2014 at 03:53 PM.
serr is offline   Reply With Quote
Old 04-02-2014, 11:05 PM   #8
osflaa
Human being with feelings
 
osflaa's Avatar
 
Join Date: Dec 2013
Location: Trondheim, Norway
Posts: 376
Default

I would go for alternative A, partly because I believe the fewer conversions, the better, partly because it's a simpler/faster way and partly because I wouldn't worry too much about quality degration in this case.
osflaa is offline   Reply With Quote
Old 04-03-2014, 02:02 AM   #9
jazznfunk
Human being with feelings
 
Join Date: Aug 2012
Location: Riga Latvia
Posts: 194
Default

Don't touch (resample) source! Reaper easy handle all standart SR's realtime. If You're paranoic (in a good way) and have mighty computer, You can put also playback resample to "EXTREME" setting, or close to. "GOOD" is enough for me. Working in high SR (aka 96 kHz) some plugins may benefit from that. I used to processing and render my master in 96 kHz and "EXTREME" resample to desired SR (44.1 or 48 usually) in render settings.

Last edited by jazznfunk; 04-03-2014 at 02:23 AM.
jazznfunk is offline   Reply With Quote
Old 04-03-2014, 03:42 AM   #10
G58
Human being with feelings
 
Join Date: Nov 2013
Posts: 44
Default

http://www.voxengo.com/product/r8brain/

useful free converter regarded as one of the least damaging; reportedly better than DAWs
G58 is offline   Reply With Quote
Old 04-04-2014, 04:56 AM   #11
Kamil
Human being with feelings
 
Kamil's Avatar
 
Join Date: Nov 2013
Location: KW, Ontario
Posts: 268
Default

I would limit the sample rate conversion to one step, much like you would dither. It is perfectly fine to have media of different sample rates / bit rates in a given project.

If you want to look at how different DAWs applications handle SRC (Sample Rate Conversion), I would check out this site.
http://src.infinitewave.ca/ - It compares them in a very understandable manner

I highly recommend SoX for SRC if you are tech-savvy (It's a commandline tool). I wish REAPER would implement it (Covered in another thread somewhere on this forum).
__________________
Ambition, Transmission, Transition.
Kamil is offline   Reply With Quote
Old 04-04-2014, 09:39 AM   #12
beingmf
Human being with feelings
 
beingmf's Avatar
 
Join Date: Jul 2007
Location: Jazz City
Posts: 5,065
Default

Quote:
Originally Posted by Kamil View Post
I highly recommend SoX for SRC if you are tech-savvy (It's a commandline tool). I wish REAPER would implement it (Covered in another thread somewhere on this forum).
SoX is available as an add-on for foobar, too. That's how I use it, pretty simple
__________________
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 04-04-2014, 10:30 AM   #13
Kamil
Human being with feelings
 
Kamil's Avatar
 
Join Date: Nov 2013
Location: KW, Ontario
Posts: 268
Default

Quote:
Originally Posted by beingmf View Post
SoX is available as an add-on for foobar, too. That's how I use it, pretty simple
Oh really? Well darn. I'm unfamiliar with foobar but I will look into it. I've just been writing scripts to do batch conversions.
__________________
Ambition, Transmission, Transition.
Kamil is offline   Reply With Quote
Old 12-11-2014, 04:39 AM   #14
huberkinky
Human being with feelings
 
Join Date: Feb 2014
Posts: 164
Default

Quick question.

I have ported digitally from old PTools into Reaper song with 44.1 kH/24 bit.

Than I've converted them into 96 k.

The question is: does Reaper leaves an origins file when glued and making a new. Or he redoing an original file and old 44.1 khz is gone.

The reason I'm asking is, I have many songs on PTools and I have to transfer them to Reaper. All songs are in 44.1 khz/24 bits.

When I'm runni9ng digitally connected Reaper with Ptools, the Reaper is switching DAW as well to 44.1. And that is causing some crashes.

Help please?
huberkinky is offline   Reply With Quote
Old 12-11-2014, 08:31 AM   #15
serr
Human being with feelings
 
Join Date: Sep 2010
Posts: 12,562
Default

Quote:
Originally Posted by frets View Post
I have a song with many tracks in reaper.
There is a mixture of 44.1khz and 48khz tracks within the song.
I eventually want to release at 44.1khz.

What is better:

A: just render the song in reaper at 44.1khz.
[ie. the resampling will be done on each track individually and then rendered together]

B: Render the song at 48khz. Then open that mixdown in a new reaper project, and render to 44.1khz.
[ie. run the resampling algorithm once on the full mix]

Why?
The 48k to 44.1k conversion is in fact the most "artifact-y" (since it starts as lower def to begin with). For perspective, it's not quite as destructive as reducing your wordlength from 24 bit to 16 bit (unless using a very very old sample rate conversion process).

For any production, always keep the sample rate untouched throughout the production/mix. The only exception should be a stray bit of content recorded at a different rate. This includes the scenario where 99% of the project is lower def and one file comes along at a higher rate.

Your final full quality master will be 24 bit 48k (And it's just fine to release at 48k if that's what you have. Don't upsample it to 96k.)
Then make the reduction to 44.1k for the portable formats like CD that require it and use the best tools available for the conversions. Creating portable versions of your master should be part of the 'mastering' process.

Quote:
Originally Posted by huberkinky View Post
Quick question.

I have ported digitally from old PTools into Reaper song with 44.1 kH/24 bit.

Than I've converted them into 96 k.
If the rest of the project (or at least a good part of it) is recorded at 96k I might do that. Upsampling 44.1k to 96k is virtually lossless. This is even done intentionally in order to get better results from some noise reduction or time/pitch expansion/compression algorithms. And it really works. Point is there's no loss from this upsample like you get from specifically the 48k to 44.1k conversion.

Convert the existing 44.1k tracks to 96k. Record any new parts/overdubs at 96k. Final release format is 24 bit 96k. Any portable master reductions (to 44.1k) will be just fine.

If everything is recorded at 44.1k though, I'd leave it. Nothing wrong with releasing a 24 bit 44.1k recording. The bigger factor is still 24 bit vs. 16 bit. Don't upsample to 96k in this case. The whole point of HD is preserving that extra 5% of fidelity - you don't want to release something that masquerades as HD.

Last edited by serr; 12-11-2014 at 08:51 AM.
serr is offline   Reply With Quote
Old 12-12-2014, 02:46 AM   #16
evosilica
Human being with feelings
 
Join Date: Jan 2011
Location: Austria/Europe
Posts: 216
Default

Haven't read all the posts but this is my opinion:

- Mix at 48kHz. Some nonlinear plugins will benefit from the little extra bandwidth (less aliasing)
- Render with a few dBs of headroom. (SRC will create new peaks, that can clip)
- Convert Samplerate
- LISTEN! Preferably do a ABX test if you can hear a difference. chances are very high you can't in an ABX setting
- Do Mastering

If you wanna be anal about SRC, check out this page
http://src.infinitewave.ca/ (oops, has already been mentioned)
REAPER's EXTREME setting looks pretty bad in comparision to others. BUT, and thats a big BUT, look at the decibel scale! Most of the artefacts are at -100dB or less.
Noone is able to hear this because of masking. If you final medium is 16bit, those artefacts are below noisefloor and therefore gone anyway.
If you disagree, go to an audiophool forum and discuss this among other very important topics such as 24bit dither and speaker cable pebbles

Last edited by evosilica; 12-12-2014 at 02:53 AM.
evosilica is offline   Reply With Quote
Old 12-14-2014, 04:57 AM   #17
bennetng
Human being with feelings
 
Join Date: Jun 2010
Posts: 264
Default

Quote:
Originally Posted by evosilica View Post
REAPER's EXTREME setting looks pretty bad in comparision to others.
Still much better than more expensive DAWs like Steinberg's and SADiE

SoX is god, even when compared with most commercial products.
bennetng 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 05:30 AM.


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