Mixing @ 96k for 44.1 (What file sounds better?)

[rhythminmind]

Before i start working on my next group of tracks.
I wanted to test if running my DAW/DSP/ & Plugins @ 96k would have any audible improvement when my final master will be 44.1.
EQ/filter curves,aliasing & whatnot perform differently at 96k so i made a quick test this morning comparing a multitrack ITB example mixed @ 96 to 44.1. & 44.1 to 44.1.

Both sessions use the exact same processing & settings. The only difference is one was running @ 96k with upsampled source tracks.

The session contained
11 drum tracks.
1 virtual EP
1 synth pad
SSL Duende - for all EQ/compression
URS Saturation

FYI - This is just some BS noodling & a rough test mix.

What file do you think sounds better?

File 1.wav = 96000 mix to 44100

File 2.wav = 44100 mix to 44100

I'm curious to see if the benefits of 96k outweigh the upsampling artifacts.

Edit ---------------------------------
Do to the confusion some are having with the files having different start times i just created new files from a different mix. These files are of equal length.
This mix also only contains EQ , Compression, & saturation processing. So there isn't any verbs, FX or vsti's being re-rendered.
These 2 files will show the mix differences of working at 96.

Drum1.wav = 96000 mix to 441

Drum2.wav = 44100 mix to 44100

[willerichey]

I think file 1 sounds more detailed, open and clear that was the 96 one.

[rhythminmind]

I'll tell whats what pretty soon. I would like to blind opinions 1st.

[danielbest]

1 sounds better, it has more HF content...

[basskater87]

hmmmm I like 1 better

[JordanA]

I also think 1 sounds the best. To me the bass end sound more defined and tuned and high end sounds more open on 1.

[AlphaDingo]

File 1.

[6strings]

I'm distracted from the get-go. The very first brush hit sounds completely different between the two files. The snare ring is totally cut or filtered out in file 1 and is generally filtered out throughout the performance. Why is that?

IMO, the differences in the snare sound outweigh anything else, so (in this case) it depends what you want your primary rhythm instrument, the snare, to sound like... I think the ring is a little much for my taste in #2 so I'll go with #1.

[UnDeFiNeD]

Two different performances, so impossible to directly compare.
quality-wise they sound identical.

Alex

[rhythminmind]

Originally Posted by UnDeFiNeD

Two different performances, so impossible to directly compare.
quality-wise they sound identical.

Alex

same performance different file lengths. File 2 just starts earlier.

[rhythminmind]

Originally Posted by 6strings

The snare ring is totally cut or filtered out in file 1 and is generally filtered out throughout the performance. Why is that?

The only differences can be the plugin response @ 96k. Exact same settings.

[UnDeFiNeD]

Originally Posted by rhythminmind

same performance different file lengths. File 2 just starts earlier.

You are right. I misaligned, I'm sorry.

Still no significant difference, other than the effects on the ePiano that seem to do different things in both takes (when I test, I allign both to be able to switch with my keyboard and my eyes closed so I am not influenced by what I see)

At first I also thought the bass on 1. had more impact, but after relistening over and over, they sound 99.99% the same. There is some difference, but not enough to decide that one is better than the other.
Thank you for testing tho, important things to test, decide and then get on with our bussiness with one less something to worry about

All ime of course...

Alex

[rhythminmind]

Well "hear" it is.


file 1 = 96k mix
file 2 = 44,1 mix

[rhythminmind]

Do to the confusion some are having with the files having different start times i just created new files from a different mix. These files are of equal length.
This mix also only contains EQ , Compression, & saturation processing. So there isn't any verbs, FX or vsti's being re-rendered.
These 2 files will show the mix differences of working at 96.

Drum1.wav

Drum2.wav

[camerondye]

Rythminmind...do you hear a difference in the two??? I get the impression that you do...what is the bonus of upsampling instead of just recording the initial file in 96k??? I'm reading this again and if you detect any smartassness, I really don't mean it.

I am actually considering the switch to 96k and just haven't yet, I use UAD plugs and I've heard they upsample to 96k anyway. They have such great sound and are an important part of my sound and it would be awesome if I could get better sound without having to buy anything for once. What I don't understand is if they work at 96k anyway, why are there less instances when you use 96k in the first place? I would think you would get more if anything because they don't have to upsample.

I just listened to the new drum1 and drum2 files and drum2 seems to be louder or compressed more, is that part of the process? I keep going back and forth on which one I like more....but it is probably drum2.
cam

[chris carter]

There is a greater difference in these files than just the sample rate. I mean, if you look at the waveforms, they are wildly different. I suspect there's more going on here than just the sample rate. Like, the plug-in instrument (I'm assuming that's what it is) is actually PLAYING DIFFERENTLY. In fact, in the first set (1.wav, 2.wav) they are actually playing different parts!

In other words, I don't think this is a test between 96kHz and 44.1kHz. I think it's a test of that plus some other junk. I think if you want to test 96/44.1 you should start with a source that's 100% controllable. Like external sources being recorded that won't play differently at different sample rates as opposed to these VST instruments which clearly are plalying two different things each time. Maybe there's some 'randomize' feature in the instruments or somethign, but it's definitely NOT the same performance each time.

[rhythminmind]

Originally Posted by chris carter

There is a greater difference in these files than just the sample rate. I mean, if you look at the waveforms, they are wildly different. I suspect there's more going on here than just the sample rate. Like, the plug-in instrument (I'm assuming that's what it is) is actually PLAYING DIFFERENTLY. In fact, in the first set (1.wav, 2.wav) they are actually playing different parts!

In other words, I don't think this is a test between 96kHz and 44.1kHz. I think it's a test of that plus some other junk. I think if you want to test 96/44.1 you should start with a source that's 100% controllable. Like external sources being recorded that won't play differently at different sample rates as opposed to these VST instruments which clearly are plalying two different things each time. Maybe there's some 'randomize' feature in the instruments or somethign, but it's definitely NOT the same performance each time.

You didn't read the post completely. the 1st file set has different start times but can be synced. I posted a 2nd example as well. Look for the "Edit" in the OP

[rhythminmind]

Originally Posted by camerondye

Rythminmind...do you hear a difference in the two??? I get the impression that you do...what is the bonus of upsampling instead of just recording the initial file in 96k??? I'm reading this again and if you detect any smartassness, I really don't mean it.

I am actually considering the switch to 96k and just haven't yet, I use UAD plugs and I've heard they upsample to 96k anyway. They have such great sound and are an important part of my sound and it would be awesome if I could get better sound without having to buy anything for once. What I don't understand is if they work at 96k anyway, why are there less instances when you use 96k in the first place? I would think you would get more if anything because they don't have to upsample.

I just listened to the new drum1 and drum2 files and drum2 seems to be louder or compressed more, is that part of the process? I keep going back and forth on which one I like more....but it is probably drum2.
cam

I'll quote this from another post.

It's all ready been said (thanks Dan) but I would like to simplify with a basic summary in lame man terms (layman's). Apply this info for your tasks at hand.

Let me bring up some technical facts of what happens @ different samplerates. There really isn’t any voodoo involved.
You have to break it down into stages.

Part One.
SampleRate in regards to your hardware(Your A/D D/A’s)
The available digital frequency response = 1/2 the sample rate.
So keep this chart in mind
44,100hz = 22.050hz
48,000hz = 24,000hz
88,200hz = 44,100hz
96,000hz = 48,000hz
Now lets say we want to have clean 20-20,000hz recording capabilities. This is what most A/D D/A’s strive for.
All A/D’s use filters on there inputs.
If you are recording @ 44,100hz and are trying to get a clean 20-20,000hz the filter only has 2.05k of bandwidth to do the cutoff. This is extremely tight & near imposable to do transparently. Most A/D’s will cut into the audible realm @ 44,100.
Changing to a higher samplerate allows the A/D D/A’s filter to have a more graceful cutoff slope & get out of the audible range.
44.1k gives the filter 2.05k of bandwidth
48k gives the filter 4k of bandwidth
88.2k gives the filter 24,100k of bandwidth
96k gives the filter 28,000k of bandwidth
This is the main reason to record with higher rates @ the hardware level.
Part Two
SampleRate in regards to processing DSP/Plugins

With added frequency bandwidth you get a few benefits. More natural EQ/filter slopes, less audible aliasing/distortion.
EQ Example
filter slopes extend out based on the Q&gain setting. If you boost the high end @ 44,100 the slope can slam into the 22.050hz limit & not be able to extend out into it’s desired range. = harsh slope. With higher rates filter slopes can extend naturally
Here are some charts of pink noise being played thru an digital EQ with a 16k boost @ 48K & 96K with the exact same settings.
48k=



96k=


Forgive my grammar.

[miqer]

It is clear on my monitors that number 2 has more distortion, a bit more compression in a bad way. The 1ste (96khz) has a nicer rounder sound.

So really do the plugs sound better at 96, and then at 44?

Perhaps cause it's distortion I'm hearing, it's the URS saturation plug that behaves different.

I can't imagine the ssl duende distorting more at 44 than at 96.

Are you sure something is not clipping at the 44.1 version, plug setting wise? Did you convert the drums from 44 to 96? How?


m.

[Soulelectronics]

Drum 1 has more depth

[rhythminmind]

Originally Posted by miqer

It is clear on my monitors that number 2 has more distortion, a bit more compression in a bad way. The 1ste (96khz) has a nicer rounder sound.

So really do the plugs sound better at 96, and then at 44?

Perhaps cause it's distortion I'm hearing, it's the URS saturation plug that behaves different.

I can't imagine the ssl duende distorting more at 44 than at 96.

Are you sure something is not clipping at the 44.1 version, plug setting wise? Did you convert the drums from 44 to 96? How?


m.

Yeah the URS sat is responding drastically different. Surprised me.
& i used sonar 8PE to upsameple & mixed the raw drum takes at 96 .