Upsampling from 44.1 to 96khz for mixing, a mistake?

chymer · 13th October 2006, 05:50 AM

Hey Slutz,
I have a question and I hope someone can help me.
Ive recently finished recording/producing a song at 24bit/44.1khz.
my client nis having it mixed in LA (im in Australia) and he swears that the mixer will want the files at 24bit/96khz.
Now, I told him that I could upsample them to 96khz but there would be no gain in quality, just file size:-)
Anyway, I think he might be insecure about "impressing" the LA cats, so I said I would upsample the files for mixing.
It got me thinking though, am i going to be compremising the quality of my files during the upsampling process? I would probably be doing it in wavelab as a batch process after exporting my files from my DAW.
I have never done it before and although I know Im not gaining quality, am I in fact in danger of damaging my files?
here are my concerns.
1- the files will sound crunchy or distort
2- the phase integrity on my drum files and the samples augmenting them will be lost.
Um, thats about it really.
Im happy to send the guy a 24/44 session, but maybe there is a benefit mixing at 96khz in terms of plugin quality and headroom and the like.....
Anyway, some advice and real world experiences would be greatly appreciated.
BTW, Ive heard the voxengo R8 brain is a good plug for upsampling.
cheers
Chymer

mixerguy · 13th October 2006, 05:53 AM

firstly, ask the mixer

secondly, if you haven't hired a mixer, hire ME!

but my instinct (not knowing more details) is don't bother.

gsilbers · 13th October 2006, 06:20 AM

are u in pro tools?

if so , you can just create a new session at 96k 24 and import session data making sure u select "copy audio files. "

if you are just sending audio files then make a bounce of each track , all from the same start and finish. you dont want the engineer and the artist trying to figure out the sync just to find out later that that bass was for the 3rd chorus or something. and bounce them at 96k 24.

but there wont be a difference whatso ever on sound quality on you tracks recorded at 44.1.
the idea is that when the engineer is mixing he will problably run the audio out to external gear and record it back into pro tools or DAT etc. then do a downsampling using a good dithering plug or recorded to a master at 44.1.

but good post before. email the engineer and talk to him directly.

but if u have pro tools just send the pro tools session taking out extra playlists and consolidating the tracks from start to finish. then doing what i saidbefore about importing session data at 96. and label it clearly.

Mike Chav · 13th October 2006, 06:26 AM

this will most definatley provide a beter end result when you print back into Pt at 96 k

chymer · 13th October 2006, 06:34 AM

Ok, thanks for the comments.
I dont use pro-tools, but all the files will be consolidated from bar 1 and will be 24bit/44.1.
I told the client that I wouldnt bother upsampling but i dont think he wants the mixer to think we recorded the song at 24bit/44.1.
I think its silly but I told him to ask the mixer what he would prefer but the client doesnt want the mixer to suspect we recorded it at 44.1.
yeah, so anyway, my question wasnt really answered, will I be compromising the quality of my files (I know I wont be gaining quality) I wanna know if the process of upsampling is a no brainer with no side effects or is there a possibility of my files being ruined. If the answer is "do it and listen and decide for yourself" then thats what I will do.
Thanks
Chymer

chymer · 13th October 2006, 06:36 AM

Yeah, apparently everyone in LA works at 96k.....
Chymer

Rep · 13th October 2006, 06:47 AM

If You have a 16 bit 44.1 mhz file that you are exporting

Going up to 24 bit 88.2 file ... you are Just adding 000000"s to you files

I will Not / Can Not ... INCREASE the fidelity,
It CAN NOT ADD to something ... that is Not there in the First place ...

chymer · 13th October 2006, 07:50 AM

Quote:

Originally Posted by Rep

If You have a 16 bit 44.1 mhz file that you are exporting

Going up to 24 bit 88.2 file ... you are Just adding 000000"s to you files

I will Not / Can Not ... INCREASE the fidelity,
It CAN NOT ADD to something ... that is Not there in the First place ...

Gee thanks man, I think I all ready made it clear that I know this, but thanks for clarifying it.
Im actually wondering if the quality is at risk of being degraded if upsampling.
I dont know if you read my post but the session is 24bit/44.1khz and I want to only upsample to 96khz, I dont nead to up the bandwidth as well, just the sample rate.
thanks
chymer

mixerguy · 13th October 2006, 08:03 AM

i am NOT impressed from what i know of sample rate conversion in ProTools

(i have done careful tests)

so - based on that - going from 44.1 to 96 k you would LOSE fidelity.

chymer · 13th October 2006, 08:27 AM

OMG an actual answer form experience, maybe I should get you to mix the song:-)
Thanks man, I will not touch the files , If the mixer wants to import them as 96 then good on him.
Chymer

WildCowboys · 16th October 2006, 08:43 AM

upsampling will usually not degrade your quality. downsampling does degrade. if you wanna be sure just upsample to 88,2 instead of 96, that means the interpolation just puts one sample inbetween two of your original samples (if you recorded at 44,1kHz). no change of your original sample-positions whatsoever, no recalculation, just double the samplingrate. this process CANNOT degrade the soundquality. the mixer guy will be satisfied with 88,2 to work with just as with 96kHz for sure.
I think the whole point is letting the mixer and client NOT know that you recorded "only" 44,1kHz, right? Otherwise the mixer could easily convert the files himself.

and yes, the mix might sound a bit better when done at double sampling rate, even though the quality of your source material will not improve by upsamling, the processing during the mix benefits from it, although not by a big margin. The mixers skillz are about a world more important to the final sound than the high samplingrate.

Rock on!
Pat

numrologst · 16th October 2006, 10:14 AM

it's been my experience that an upsample will degrade the quality of audio. I once was mixing a seesion recorded at 48k and then upsampled to 96k... I got a hold of the 96k files and there was werid ghost frequencies, tons of weird high frequency distortion... it was really weird...

I called the tracking engineer to get the 48k files...At this point the drive that they were on went down and the 48k files were lost from the whole session. I then had to spend a full 12 hour day of fixing all the files, noise reduction tons of low pass eq. It was a nightmare.

I've seen the same thing happen with downsamples from 192k to 48k... As well as from 88.2k down to 44k.

Now I make it practice to downsample/bit reduce exclusively with my apogee 16x's... Nothing else will suffice for me at this point

WildCowboys · 16th October 2006, 11:04 AM

Quote:

Originally Posted by numrologst

it's been my experience that an upsample will degrade the quality of audio. I once was mixing a seesion recorded at 48k and then upsampled to 96k... I got a hold of the 96k files and there was werid ghost frequencies, tons of weird high frequency distortion... it was really weird...

I called the tracking engineer to get the 48k files...At this point the drive that they were on went down and the 48k files were lost from the whole session. I then had to spend a full 12 hour day of fixing all the files, noise reduction tons of low pass eq. It was a nightmare.

I've seen the same thing happen with downsamples from 192k to 48k... As well as from 88.2k down to 44k.

Now I make it practice to downsample/bit reduce exclusively with my apogee 16x's... Nothing else will suffice for me at this point

there must have been something really wrong. this is not what happens when upsampling normally. doubling the samplingrate is the easiest, artifact-free calculation ever. you could theoretically downsample the files again and end up with exactly the same file you started with...

It seems like a matter of finding the right software to do the job, but nowadays I do not think one could go wrong with any professional software...

rock on!
Pat

DarkSky · 16th October 2006, 11:27 AM

I agree with others that upsampling will involve some loss of quality. Its effect should not be great (depends on the SRC) but conversions are best kept to a minimum. It's also true that mixing at a higher sample rate *may* give better results - but this depends on exactly what is being done and how (it could also give worse results in some situations/set-ups).

So there is a trade-off and the question is will the benefits (if any) of running the mix at 96k (or 88.1 as someone suggested) be significant and if so will it be sufficient to outweigh any loss incurred in conversion. Trouble is, you can't know this. You have access to only part of the information - you'll be able to judge how much is lost in the SRC - but the mixer is the one who should know what the extent of benefits will be to compensate.

jseveriniii · 16th October 2006, 11:31 AM

Upsampling will not in any way degrade your sound. DOWNsampling WILL degrade your audio. Period. You can't argue with science. Look it up in any pro magazine, book, audio theory course.

DarkSky · 16th October 2006, 01:16 PM

Quote:

Originally Posted by jseveriniii

You can't argue with science.

LOL. Sorry to break it to ya, but scientists are *noted* for arguing among themselves - they do it all the time. Some would say its part of what science is *about*.

But that aside, *every* time audio is converted to new binary values, rounding errors (at the very least) are introduced. Depending on the algorithm/s used to do the conversion (and different chipsets implement SRC in different ways) other errors may also be introduced - some more audible than others.

For those of us who've heard poor SRC results on upsampling, your chapter and verse of Digital Audio 101 is not very convincing. If it comes to believing something that someone wrote in a magazine or believing the evidence of one's own ears, I'll go with the ears. YMMV.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Upsampling for mastering	pingu	Mastering forum	9	12th April 2006 12:06 PM
Upsampling	ttauri	So much gear, so little time!	1	14th March 2005 03:49 AM
Upsampling 44.1 to 48 for video?	bdunard	So much gear, so little time!	7	20th February 2005 07:15 PM
Pro Tools: Upsampling From 48k to 96k Quality Loss?	Perspectrum	Music computers	1	25th February 2004 12:22 PM
I must have made a mistake.......	hollywood_steve	Remote Possibilities in Acoustic Music & Location Recording	10	1st October 2003 11:21 PM

13th October 2006, 05:50 AM	#1
chymer Lives for gear Join Date: Oct 2005 Location: Sydney Posts: 520	Upsampling from 44.1 to 96khz for mixing, a mistake? Hey Slutz, I have a question and I hope someone can help me. Ive recently finished recording/producing a song at 24bit/44.1khz. my client nis having it mixed in LA (im in Australia) and he swears that the mixer will want the files at 24bit/96khz. Now, I told him that I could upsample them to 96khz but there would be no gain in quality, just file size:-) Anyway, I think he might be insecure about "impressing" the LA cats, so I said I would upsample the files for mixing. It got me thinking though, am i going to be compremising the quality of my files during the upsampling process? I would probably be doing it in wavelab as a batch process after exporting my files from my DAW. I have never done it before and although I know Im not gaining quality, am I in fact in danger of damaging my files? here are my concerns. 1- the files will sound crunchy or distort 2- the phase integrity on my drum files and the samples augmenting them will be lost. Um, thats about it really. Im happy to send the guy a 24/44 session, but maybe there is a benefit mixing at 96khz in terms of plugin quality and headroom and the like..... Anyway, some advice and real world experiences would be greatly appreciated. BTW, Ive heard the voxengo R8 brain is a good plug for upsampling. cheers Chymer

13th October 2006, 06:26 AM	#4
Mike Chav Gear nut Join Date: Dec 2005 Posts: 83	upsampling this will most definatley provide a beter end result when you print back into Pt at 96 k

13th October 2006, 06:47 AM	#7
Rep Lives for gear Join Date: Dec 2003 Location: So-Cal Posts: 1,562	If You have a 16 bit 44.1 mhz file that you are exporting Going up to 24 bit 88.2 file ... you are Just adding 000000"s to you files I will Not / Can Not ... INCREASE the fidelity, It CAN NOT ADD to something ... that is Not there in the First place ... __________________ The only regrets We will have in Life......Are the things we Never Tried To do.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes	Rate This Thread
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode	Rate This Thread:

13th October 2006, 05:53 AM	#2
mixerguy Lives for gear Join Date: Sep 2005 Posts: 3,013	firstly, ask the mixer secondly, if you haven't hired a mixer, hire ME! but my instinct (not knowing more details) is don't bother.

13th October 2006, 06:20 AM	#3
gsilbers Lives for gear Join Date: Jan 2006 Posts: 2,361	are u in pro tools? if so , you can just create a new session at 96k 24 and import session data making sure u select "copy audio files. " if you are just sending audio files then make a bounce of each track , all from the same start and finish. you dont want the engineer and the artist trying to figure out the sync just to find out later that that bass was for the 3rd chorus or something. and bounce them at 96k 24. but there wont be a difference whatso ever on sound quality on you tracks recorded at 44.1. the idea is that when the engineer is mixing he will problably run the audio out to external gear and record it back into pro tools or DAT etc. then do a downsampling using a good dithering plug or recorded to a master at 44.1. but good post before. email the engineer and talk to him directly. but if u have pro tools just send the pro tools session taking out extra playlists and consolidating the tracks from start to finish. then doing what i saidbefore about importing session data at 96. and label it clearly.

13th October 2006, 06:34 AM	#5
chymer Lives for gear Join Date: Oct 2005 Location: Sydney Posts: 520	Ok, thanks for the comments. I dont use pro-tools, but all the files will be consolidated from bar 1 and will be 24bit/44.1. I told the client that I wouldnt bother upsampling but i dont think he wants the mixer to think we recorded the song at 24bit/44.1. I think its silly but I told him to ask the mixer what he would prefer but the client doesnt want the mixer to suspect we recorded it at 44.1. yeah, so anyway, my question wasnt really answered, will I be compromising the quality of my files (I know I wont be gaining quality) I wanna know if the process of upsampling is a no brainer with no side effects or is there a possibility of my files being ruined. If the answer is "do it and listen and decide for yourself" then thats what I will do. Thanks Chymer

13th October 2006, 06:36 AM	#6
chymer Lives for gear Join Date: Oct 2005 Location: Sydney Posts: 520	Yeah, apparently everyone in LA works at 96k..... Chymer

13th October 2006, 08:03 AM	#9
mixerguy Lives for gear Join Date: Sep 2005 Posts: 3,013	i am NOT impressed from what i know of sample rate conversion in ProTools (i have done careful tests) so - based on that - going from 44.1 to 96 k you would LOSE fidelity.

13th October 2006, 08:27 AM	#10
chymer Lives for gear Join Date: Oct 2005 Location: Sydney Posts: 520	OMG an actual answer form experience, maybe I should get you to mix the song:-) Thanks man, I will not touch the files , If the mixer wants to import them as 96 then good on him. Chymer

16th October 2006, 08:43 AM	#11
WildCowboys Lives for gear Join Date: Nov 2003 Location: VIE Posts: 1,072	upsampling will usually not degrade your quality. downsampling does degrade. if you wanna be sure just upsample to 88,2 instead of 96, that means the interpolation just puts one sample inbetween two of your original samples (if you recorded at 44,1kHz). no change of your original sample-positions whatsoever, no recalculation, just double the samplingrate. this process CANNOT degrade the soundquality. the mixer guy will be satisfied with 88,2 to work with just as with 96kHz for sure. I think the whole point is letting the mixer and client NOT know that you recorded "only" 44,1kHz, right? Otherwise the mixer could easily convert the files himself. and yes, the mix might sound a bit better when done at double sampling rate, even though the quality of your source material will not improve by upsamling, the processing during the mix benefits from it, although not by a big margin. The mixers skillz are about a world more important to the final sound than the high samplingrate. Rock on! Pat

16th October 2006, 10:14 AM	#12
numrologst Lives for gear Join Date: Sep 2005 Location: Texas Posts: 1,894	it's been my experience that an upsample will degrade the quality of audio. I once was mixing a seesion recorded at 48k and then upsampled to 96k... I got a hold of the 96k files and there was werid ghost frequencies, tons of weird high frequency distortion... it was really weird... I called the tracking engineer to get the 48k files...At this point the drive that they were on went down and the 48k files were lost from the whole session. I then had to spend a full 12 hour day of fixing all the files, noise reduction tons of low pass eq. It was a nightmare. I've seen the same thing happen with downsamples from 192k to 48k... As well as from 88.2k down to 44k. Now I make it practice to downsample/bit reduce exclusively with my apogee 16x's... Nothing else will suffice for me at this point

16th October 2006, 11:27 AM	#14
DarkSky Lives for gear Join Date: Apr 2006 Location: Melbourne, Australia Posts: 892	I agree with others that upsampling will involve some loss of quality. Its effect should not be great (depends on the SRC) but conversions are best kept to a minimum. It's also true that mixing at a higher sample rate may give better results - but this depends on exactly what is being done and how (it could also give worse results in some situations/set-ups). So there is a trade-off and the question is will the benefits (if any) of running the mix at 96k (or 88.1 as someone suggested) be significant and if so will it be sufficient to outweigh any loss incurred in conversion. Trouble is, you can't know this. You have access to only part of the information - you'll be able to judge how much is lost in the SRC - but the mixer is the one who should know what the extent of benefits will be to compensate.

16th October 2006, 11:31 AM	#15
jseveriniii Gear nut Join Date: Oct 2006 Location: Los Angeles, CA Posts: 129	Upsampling will not in any way degrade your sound. DOWNsampling WILL degrade your audio. Period. You can't argue with science. Look it up in any pro magazine, book, audio theory course.