Question regarding Bouncing and SRC

[huejahfink]

I don't use protools, but this is how I would work it out.
Where do you set the ceiling on the TC?
If you set your ceiling to 0.0 then it might not be easy to tell as the signal will just clip, so try with a lower ceiling like -1 as a test instead.

Do peaks overshoot it?

If no - excellent. All is well.

If yes - then tools is performing SRC last, so you should do your dithering as a separate step.

[Beech]

Yes, I set the ceiling on the TC to -.01db and I don't get any plugin clipping.

I'm not sure that the plugin clipping light on the TC is necessarily a sign of SRC being done last in the processing chain though ...

[Trakworx]

Originally Posted by Beech

Yes, I set the ceiling on the TC to -.01db and I don't get any plugin clipping.

I'm not sure that the plugin clipping light on the TC is necessarily a sign of SRC being done last in the processing chain though ...

First, I'd set the output ceiling to -0.3 or -0.5 to help prevent ISPs when converting to MP3, etc.

Also, you have to check the bounced 16/44 file for peak levels to see what is happening. The meter in your 24/88 session will not show you what happens "after" it. Import the bounced file into a 16/44 session. PT's "Gain" AudioSuite plug can find peak or rms levels for you.

Let us know...

[Ruairi]

Hi Daire,

Ruairi here (formerly of Killarney but now L.A.), I rented some gear from you in the past.

If you're working in Pro Tools I'd recommend making a small investment and buying Sample manager, I think it's $79. It uses the iZotope SRC which is the best I've heard and easily superior to Pro Tools IME.

I record finished mixes or masters back to a new track on PT and export at 88.2/24 or whatever the native rate is. Then SRC and dither using Sample Manager.

It also makes delivery of different formats easy - MP3/AAC for bands website etc.

Cheers,
Ruairi

[huejahfink]

Originally Posted by Beech

Yes, I set the ceiling on the TC to -.01db and I don't get any plugin clipping.

I'm not sure that the plugin clipping light on the TC is necessarily a sign of SRC being done last in the processing chain though ...

In case there is any confusion, I mean to check the file "after" export not during. Where do the peaks reach then when examining the exported file if you set the ceiling to -0.5 for example?

If it is anything like Cubase, then the SRC happens post fader + inserts, meaning that the peaks caused by the SRC would be clipping your signal if you were exporting at a fixed point bit depth. My guess would be that Tools would operate in the same way otherwise it would be doing SRC on the fly at some point before the master inserts which seems pretty unlikely.

(+1 with Ruairi about iZotope SRC by the way. Worth getting hold of that somehow.)

[Beech]

Thank you all kindly for your suggestions. @Ruairi - I do indeed remember you. I hope you are keeping well.

I have taken all of your suggestions on board and I am in the process of doing some tests now. I will report back asap.

[streaky]

Don't go near PT for SRC or dither it sounds dreadful...

another +1 for iZotope SRC & dither...

[Beech]

Well, here's what I have done:

1. Bounced the 24bit 88.2k file down with the TC limiter on but not set to dither. The TC is hardly working btw, really just bringing the level up and shaving off a couple of db from the peaks.

2. Bounced the same 24bit 88.2k file straight to 16bit 44,1 using the TC as above but this time with dither. Pro Tools doing the SRC.

3. Bounced the same 24bit 88.2k file to 24bit 44.1k without dither. Then opened a new file and imported this and bounced it to 16bit using the same TC settings as in number 2. above.

So, now I'm going to use Sample Manager to SRC and dither the file from number 1. above.

Also, since there is a 24bit 44.1k file I might as well throw that one at Sample Manager to dither it to 16bit. This will give us a total of four files to choose from.

I have burned a cd of all four using Waveburner. Even though this is acoustic music with lots of tails and decays, I know it's going to be hard to pick one of these as "better" then the rest so I'll leave it till the morning when the ears will hopefully be fresher. My gut feeling at this stage is that NOT doing the SRC and the dither in the same pass is probably the way to go.

Good night.

[huejahfink]

My suggestion would be to bounce with the limiter turned off at 24 88. Then perform SRC and keep at 24.
Now perform the limiting using the same settings to the 24 44 file, and you can dither to 16bit now if you wish.
I use Cubase as my primary DAW and I export at the native rate using 24 bit file. Then I follow by iZotope RX2 adv. So in iZotope I can perform any SRC, fades, check for minute clicks or glitches I may have missed, perform final limiting (as you can use plugins in that version) - save to 24 bit and then perform dithering to 16 in a separate step all in RX2. Works out well for me.

[Ruairi]

Originally Posted by huejahfink

My suggestion would be to bounce with the limiter turned off at 24 88. Then perform SRC and keep at 24.
Now perform the limiting using the same settings to the 24 44 file, and you can dither to 16bit now if you wish.

I wouldn't do this even in a situation where the limiter is not working hard. In my experience all peak limiters I've tried sound better at 2xFS. I can't see any advantage to the way you've described above.

Sample manager allows you to choose the order of your processes - SRC first, then dither. I use the standard iZotope SRC included and use the mBit+ dither.

I'm doing this daily as part of my mastering workflow. It sounds good.

[huejahfink]

Fair enough.

I used to do it the way you describe and I agree that limiting can sound at least different when performed at higher rates, but not necessarily better as far as I am concerned. Saturation / clipping based peak control usually does sound better at higher rates to me, but regular limiting not really. Plus most of the limiters I use have inbuilt upsampling modes which can easily be auditioned and performed on the target rate file. Granted that is different to doing the limiting prior to SRC and performing any rate conversion only once.

I like my final limiter to have complete control of the ceiling and not have to be concerned about subsequent ceiling overshoots due to SRC. That's the main advantage for me, and it really suits my workflow better.
This said, if I'm only limiting a few little stray peaks here and there I'll probably just limit prior to SRC anyway.

I nearly always audition the limiter at the original native project rate first before I SRC the non limited file. If it really sounded noticeably worse I can always perform SRC after limiting instead. Hasn't been an issue for me yet though.

Different strokes I suppose! Plus at least 90% of the mixes that come in to me arrive at 44.1 anyway - .

[Beech]

I listened to the four versions that I made last night played back from a standard cd player through the studio monitors. Observations as follows:

1. This version was SRC by Pro Tools and dithered with the TC. To be honest I thought it was nice - very nice.

2. This file was first bounced to 44.1k in Pro Tools. The resulting file was then imported to a new session and bounced to 16 bit using the TC dither. Limiting had been done at the higher sample rate as this is my preferred way of working. Overall I was not impressed with the sound, it was a little "grainy", particularly noticeable on the vocal I felt.

3. File bounced down with limiting. Sample Manager was used to SRC and dither. This was the version I ended up using for the project. However, there wasn't as big a difference between this and #1 above to these ears.

4. Bounced version using PTs SRC and SM's dither. Wasn't mad about this one.

So, to conclude, thanks to all for your help. The Sample Manager version is how I plan to proceed for the album. I think I can rest peacefully knowing that I've found the least destructive method of reducing hi rez audio to 16 bit.

[Beech]

BTW, I never did get a definitive answer to the question of whether a bounce to disc in Pro Tools does it's SRC before or after the dither plugin on the master fader. My gut feeling is that it's the former but I'd like to know for sure...

[Slug1]

Check the izotope site. They have a nice write up on this. It's SRC first, so convert to 24 bit 44.1kHz and bounce. Then that track should be dithered down to 16 bit. Dither should be done post fader. Some of the limiters have good dithering algorithms. Key is turning off the internal dithering algorithm in your daw. There's a really nice SRC conversion site that shows the aliasing and noise artifacts generated by SRC in different software including protools. Check it out.