The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Gear for sale     Latest  Trending
Sample Rate Conversion
Old 28th February 2019
  #1
Sample Rate Conversion

Converters? if so which ones?

r8brain?
rx7?
Sox?
Weiss?
CDfile?

What else is there?

What settings? When and how?

From my experience, out ond dac and into a 44.1khz adc is good. r8brain at 32 bit float is ok. but not perfect.. Is anyone getting a good detailed non 2d sounding conversion from 48khz? I hear its better to convert from 88khz. is this right?

DF
Old 28th February 2019
  #2
I know you have Cubase 10 from another thread, albeit with some issues...

Cubase 10 implemented SoX SRC which is basically as good as it gets. Convert with 10 and work with 9.5.

You don't have to worry about converting from a specific sample rate down/up to another. Just convert to the sample rate you would like to work in.
Old 28th February 2019
  #3
Lives for gear
 
Trakworx's Avatar
Quote:
Originally Posted by Disease Factory View Post
From my experience, out ond dac and into a 44.1khz adc is good.
I've found that software SRC changes the sound less than a trip through DA/AD.

I tested by SRCing a 24-44.1 wave to 24-96 and then back to 24-44.1, then performing a null test vs the original. They nulled, which proved to me that the files were identical even after 2 SRCs. This was with nothing but Pro Tools Tweak Head SRC. I stopped worrying about SRC after that. I suggest performing some null tests with whatever options you have and going with whatever nulls best.
Old 28th February 2019
  #4
Lives for gear
FinalCD? (Windows only...)
Old 28th February 2019
  #5
Gear Nut
The most cleanest, top quality, linear phase, full range (steepest cut at 22kHz), without distortions, aliasing and noise:
FinalCD (sharp), converts from any sample rate, but only to 44.1
Dbpoweramp
SoX (incl. in some daws, converting software etc, as well as command line SoX)
The Brick (mac only)
RX
Foobar2000 latest version includes best algos as well

Also very good quality:
Weiss Saracon
Voxengo R8brain pro/free
Adobe Audition newest versions
EZ CD audio converter (since 8.0.6/8.0.7 and higher it does use very good quality src, dithering and so on)
Secret Rabbit Code


Use SRC before dithering to 16 bit when limitered/mastered file(s) is/are ready to be bit reduced before CD. To avoid changes in peaks and true peaks, and avoid clipping, the better way is also src before limiting/clipping, but ceiling should be below 0dB, dither is the last in the chain.
Use SRC if your project 96, for example, and samples 44 - so you need to convert your samples to 96, because daws mostly do it worse than special SRC.

Also, it is not a problem to src from 44 to 48 or vice versa. Yes, some daws or other software still do bad conversions either from 44 to 48, or 44 to 88 etc. But listed above are all high-highest-top quality without those math problems.

Most answers are here:
src.infinitewave.ca/

Compare all these expensive "pro" DAWs (ProTools, Sadie, Motu, pyramix...) vs cheapest/free top quality stuff (finalCD, sox, secret rabbit code, aimp player, reaper)...
And you will see old fart hyped stuff converts slightly worse or even bad, so some mastering grade elements of those pro dAWs must be reconsidered, reworked from groundup
Old 28th February 2019
  #6
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
Weiss Saracon remains the most consistantly "best" I've heard. If I have hours to tweak RX7 back and forth I can sometimes equal it but there is no one "best" setting I've ever found. FinalCD is consistantly the second best I've found.
Old 1st March 2019
  #7
Quote:
Originally Posted by Slogun View Post
I know you have Cubase 10 from another thread, albeit with some issues...

Cubase 10 implemented SoX SRC which is basically as good as it gets. Convert with 10 and work with 9.5.

You don't have to worry about converting from a specific sample rate down/up to another. Just convert to the sample rate you would like to work in.
I have done this, the results were very ****ty, maybe because i bounced to 16 and 24, ill try 32 bit float, and see what the results are, so sox is running in real time? So the bounce is the sox src? or do i have to convert in the pool? is there setttings i can control?

DF
Old 1st March 2019
  #8
What i learned with r8brain is that if i spit out my file with -0.4 the conversion sounds good hq 32 float. its almost perfect.. So in cubase 10, i wonder on your master fader, you would have to do this same thing.

DF
Old 1st March 2019
  #9
Quote:
Originally Posted by Disease Factory View Post
I have done this, the results were very ****ty, maybe because i bounced to 16 and 24, ill try 32 bit float, and see what the results are, so sox is running in real time? So the bounce is the sox src? or do i have to convert in the pool? is there setttings i can control?

DF
If the result sounded ****ty in C10 I'm inclined to say there's something else messing it up in the system.
As you can see by comparison different SRC's in the link Stereo Flux provided, Cubase 10 (SoX algorithm) is top notch and very well regarded out there.

And yes, I suggest you do the comparison/listening test without any bit truncating.

You don't have to do anything in Cubase to get the SoX SRC. If you have a project at for instance 44.1kHz and you import/drag audio files into the project in a different sample rate, then you get the dialog box saying that the files needs to be SRC'ed. Click Ok and it's done. It's no separate plugin or running in real time as the audio files are converted before they end up in the project window.
Old 1st March 2019
  #10
Deleted 691ca21
Guest
FinalCD for all downsampling from 96 to 44.1, and RX7 Advanced for everything else.

I used to use R8Brain Pro about a decade ago, but the above mentioned are both huge steps up in terms of sound quality, and FinalCD is free, so I would highly recommend checking that out.

I use the SoX resampler in FB2K for realtime SRC for pleasure listening, if needed.
Old 1st March 2019
  #11
Lives for gear
 

Verified Member
Every audio engineer should own Izotope RX.
Old 1st March 2019
  #12
FinalCD and Saracon here
Old 2nd March 2019
  #13
Quote:
Originally Posted by Joe_caithness View Post
Every audio engineer should own Izotope RX.
i demo'd it and honestly r8brain pro if you output your file at -0.40 it sounded better to me than rx7.

DF
Old 2nd March 2019
  #14
Lives for gear
 
Jerry Tubb's Avatar
 

Verified Member
most of the current ones are really good now, even Pro Tools Tweak Head is quite acceptable.

the task is not to use an older SRC with artifacts.

best, jt
Old 2nd March 2019
  #15
Lives for gear
 
Robo's Avatar
Quote:
Originally Posted by Trakworx View Post
I've found that software SRC changes the sound less than a trip through DA/AD.

I tested by SRCing a 24-44.1 wave to 24-96 and then back to 24-44.1, then performing a null test vs the original. They nulled, which proved to me that the files were identical even after 2 SRCs. This was with nothing but Pro Tools Tweak Head SRC. I stopped worrying about SRC after that. I suggest performing some null tests with whatever options you have and going with whatever nulls best.
A little off topic but:

Just be careful with words like 'identical' when talking about null testing.

Null testing is not an 'it is' or 'it isn't' type of thing.

Signals can null all the way to -300-and-something dB, which is a perfect null, or more precisely, a duplicate file.

But the human ear can perceive something to be 'silent' at only -60dB/-70dB or so.

Be careful when stating something 'nulls', as it is more often than not a degree of nulling, not an absolute thing.

Cheers
Old 2nd March 2019
  #16
Lives for gear
 
Trakworx's Avatar
Quote:
Originally Posted by Robo View Post
A little off topic but:

Just be careful with words like 'identical' when talking about null testing.

Null testing is not an 'it is' or 'it isn't' type of thing.

Signals can null all the way to -300-and-something dB, which is a perfect null, or more precisely, a duplicate file.

But the human ear can perceive something to be 'silent' at only -60dB/-70dB or so.

Be careful when stating something 'nulls', as it is more often than not a degree of nulling, not an absolute thing.

Cheers
Yes of course you are technically correct. I really didn't want to explain all of that when making my point so I just used the shorthand "they nulled". In fact there was a tiny remainder that showed up on an FFT above 22k and below -80dB. There was nothing else visible on the FFT and nothing audible with my monitors cranked 100% which would have equaled over 120dBspl without the null test in place. I don't know, is it OK to just call than a null for practical purposes? Or do I have to give all the technical details every time I use the word "null"? Is there another word I could use that would work better? A NILL?

"They nilled". I like it!
Old 3rd March 2019
  #17
Null is pointless, my ears are what matters. and i have been able to tell a diff with my ears even when things null. conversion esp, because the nyquest filter does effect the sound, i can hear it, some cant.. so whatever.. not sure if it happens after the dac, so even if it nulls in a computer your converter maybe effecting the sound.

DF

Last edited by Disease Factory; 4th March 2019 at 08:21 AM..
Old 3rd March 2019
  #18
Gear Nut
I don't think you hear changes that excellent SRC doesnt produce. If there are no distortions or phase shifts and so on, it will not bring any to sound. Only pre and post ringing at 22kHz or 22.05kHz. They are short, ultrasonic and not loud, so no worry. Even spectrogram shows that rx, finalcd don't introduce anything to sound, weiss saracon which is mastering grade introduces more aliasing than rx, sox, finalcd, dbpoweramp (dbpoweramp tests at src.infinitewave.ca/ were done by me), which dont do changes in the sound.
If you hear obvious changes in sound, either you use very bad old src, or bad settings, or it is just a nature of your DAC to play same sound at 96 or 44 differently.

Test 2
Can you make same recording with same settings and src this way:
Convert from your sample rate into 44.1, then this 44.1khz to 96, then 96 to 88.2, then 88.2 to 192 and so on. One file rererereconverted 20 times or more, then convert this rerereconverted file to your first sample rate you worked first. So, you have 2 files with same sample rates: original source file with sample rate (exported from DAW, not converted) , and rerererereconverted file with same sample rate as the first one.

Now compare)
Old 3rd March 2019
  #19
Lives for gear
 
Robo's Avatar
Quote:
Originally Posted by Trakworx View Post
Yes of course you are technically correct. I really didn't want to explain all of that when making my point so I just used the shorthand "they nulled". In fact there was a tiny remainder that showed up on an FFT above 22k and below -80dB. There was nothing else visible on the FFT and nothing audible with my monitors cranked 100% which would have equaled over 120dBspl without the null test in place. I don't know, is it OK to just call than a null for practical purposes? Or do I have to give all the technical details every time I use the word "null"? Is there another word I could use that would work better? A NILL?

"They nilled". I like it!
I like NILL too :D

I would probably say 'nulls to -xdB' and then people can make up their own mind about it.

A -80dB null sounds good, no two ways about it, nothing in the speakers, but what is in that extra 200 and something dB? This could be the last 10% that we are always trying to catch, subtle imaging cues, realism, detail.

By the way I don't mean to imply the SRC you used is not good, I can believe the difference is hardly noticable in the scheme of things, and that it sounds great.

Cheers
Old 3rd March 2019
  #20
Lives for gear
 
Trakworx's Avatar
Quote:
Originally Posted by Stereo Flux View Post
...or it is just a nature of your DAC to play same sound at 96 or 44 differently.
This.

A lot of opinions floating around about the sound of various sample rates and SRC can be attributed to DAC designs that implement filtering and clocking differently at different sample rates. That's why an apples to apples comparison is needed - listening at the same SR. I'm not saying that the OP didn't do that but some people don't and then the waters get muddied.

Also, if you're going to rely on just your ears then BLIND listening tests are extremely important. Expectation bias and confirmation bias are well documented and if you're a human being they affect you too. I learned it the hard way long ago. Give people 2 glasses of the same wine and tell them one costs more and they'll think that one tastes better (just one example out of many). It also matters what you're looking at when you listen. Different UIs affect how we hear. If those effects aren't controlled for in a good blind test then listening tests are of very limited value.

Give me a solid NILL test any day!
Old 3rd March 2019
  #21
Lives for gear
Quote:
Originally Posted by Disease Factory View Post
Null is pointless, my ears are what matters. and i have been able to tell a diff with my ears even when things null.
You can hear a difference when there is no difference. That is really quite an achievement!
Old 3rd March 2019
  #22
You guys should be carefull with the words you use.
Either it nulls or it doesn't but there is no half way.
If the remaining signal is -80dbfs, it doesn't null, if it's -120dBfs, it doesn't null (by a very tiny amount)
When it nulls, there is nothing remaining, the right way to check that is with a bit counter like Bitter << Stillwell Audio
Old 3rd March 2019
  #23
Lives for gear
 
Trakworx's Avatar
Sure, though I was talking about < -80 and above 22k.

It's not a null but it's a damn strong NILL!
Old 3rd March 2019
  #24
Gear Nut
-80dB is not so much difference, about subtle...
-120dB is very very invisible.
Old 4th March 2019
  #25
DAC can color your sound after and the null test will mean nothing.. converters sound different at differante khz. So this is why null tests can geat my %$%@$%@$%@[email protected]%

Last edited by Disease Factory; 4th March 2019 at 03:02 PM..
Old 4th March 2019
  #26
Lives for gear
 

I think there is a grave misunderstanding of what the results of a NULL or DIFFERENCE test yield.

They are not just 'residue' existing down in some defined noise-floor ... it represent the 'difference' between two audio sources.
Old 4th March 2019
  #27
Lives for gear
 

Yeah, a decent chain with a bunch of tubes and a kilos worth of tranny will null down to -80db with a solid state counterpart.
Old 4th March 2019
  #28
Lives for gear
 
Trakworx's Avatar
Quote:
Originally Posted by Disease Factory View Post
DAC can color your sound after and the null test will mean nothing.. converters sound different at differante khz. So this is why null tests can geat my %$%@$%@$%@[email protected]%
Eloquent.

For the benefit of anyone reading who may not be familiar with null tests I'll just point out that when ITB null testing 2 audio files that were software SRCd you are playing them both at the same sample rate through the same DAC, which removes the DAC from the equation entirely.

The problem of DACs sounding different at different SRs is real, but it's an entirely different issue from SRC.
Old 4th March 2019
  #29
Lives for gear
Quote:
Originally Posted by Disease Factory View Post
DAC can color your sound after and the null test will mean nothing.
Please explain how a DAC is going to color identical files differently.
Old 4th March 2019
  #30
Lives for gear
 
Trakworx's Avatar
Quote:
Originally Posted by Timesaver800W View Post
Yeah, a decent chain with a bunch of tubes and a kilos worth of tranny will null down to -80db with a solid state counterpart.
If you're just talking about the noise floor then I would agree, but when running signal through a bunch of tubes and transformers I routinely get added harmonics far far above -80dB.
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump