output ceiling and sample rate conversion

[Swell]

Hello everyone,

I used to bounce my masters into Pro Tools and making all down conversion during the bounce.
Knowing I could get better sample rate conversion than Digidesign's, I'm testing Sample Manager.

Here's my issue: I have a brickwall limiter at the end of my chain seted at -0.3dBFs. I bounce without conversion meaning in this case, 24_88,2
Then I load the bounce into Sample Manager and get down to 44.1
At this stage already my master clips. It announces maximum volume at+0.2. That is 0.5 dB higher than my brickwall settings.

Is it normal ?? Should I bounce with more marge ? Should I brickwall limit after sample rate conversion and for example use L2 with dither ?

How do you guys do this please ? Thanks.

Swell

[Alexey Lukin]

Increase in peak levels during SRC is normal. You should either leave more margin or render to 32-bit float format and apply peak limiter at the target sampling rate.

[wado1942]

Pro Tools won't do floating point as far as I know. However, you can save the limiter till the last stage and have an increase in aliasing or you can set the ceiling of the limiter lower.

[dave-G]

I'm not sure if there's any SRC besides the one in Peak 6 that ends up as 'gain neutral', preventing or limiting clipping in the process (is there?). It's a slow one, but it's a good SRC anyway, and that's a very handy feature.

I was told that iZotope was considering adding a similar feature for offline processing, and that if not, Audiofile Engineering may consider adding the feature themselves in some form to Sample Manager and Wave Editor (both of which use the iZotope SRC).

-dave

[Greg Reierson]

Originally Posted by Alexey Lukin

Increase in peak levels during SRC is normal. You should either leave more margin or render to 32-bit float format and apply peak limiter at the target sampling rate.

That's the way I do it. Do 95% of your limiting at the higher rate, then SRC, then a final safety limiter to catch last few tenths.


GR

[Swell]

Greg, Dave, Wado, Alexey,

Thank you for your advise. I find in Greg's "way" an interesting aspect:
limiting "artistically" at a rate not making the music suffer, then re-importing the 44.1kHz file and with a last limiter getting it louder if the client request it...
Anyone having a similar method ?

About Sample Manager and Barbabatch...any preferences ?? I like the fact that
Sample manager is powered by Izotop. There dither seems more relyable than Barbabatch's.

Thank for the exhange !

Swell

[Table Of Tone]

You will probably find that down sampling a pair of 24 bit, left and right mono files, will raise the ceiling much less.
You can do this after the limiter but before the dither.
Check it out but I think you will only see a raise of 0.1db, if you do the above on most of the better sample rate converters.
You can then go ahead and dither you audio to 16 bit, at the same time bringing your DAW output down 0.1db (before the dither!)

The results will sound closer to using two puters, one for playback (88.2k/24 bit) and one for capture (44.1k/16 bit), dithering to 16 bit on the AD converter that you are using for capture after the outboard.

Hope this helps
Give it a go

[Jenz]

I usually set the output to -0.2dB before SRC, and i've never heard a problem, even if it does indicate clipping at some point. If you re-visit your master it's a pain to first limit safely, then SRC, and then limit again. At least if the SRC is offline, and the rest in real time.

[Lagerfeldt]

Originally Posted by Table Of Tone

You will probably find that down sampling a pair of 24 bit, left and right mono files, will raise the ceiling much less.

Be careful doing that. I've seen and heard problems when SRC'ing split mono files as opposed to SRC'ing an interleaved file.

I even did some testing and came to the conclusion that it's safer not to SRC on split files.

[Swell]

Then if you find dangerous to SRC split files, would you trust Pro Tools doing the interleaved bounce ? I'm mean letting PT boncing and converting the split files into interleaved ?

Thanks !

Swell

[Table Of Tone]

Originally Posted by Lagerfeldt

Be careful doing that. I've seen and heard problems when SRC'ing split mono files as opposed to SRC'ing an interleaved file.

I even did some testing and came to the conclusion that it's safer not to SRC on split files.

What happened?
Did you get phasing issues?
What SRC and what DAW did you use?

If I master on only one puter and the playback is digital and upwards of 48k, I'll use soundblade in a zinman loop to capture (still at 24 bit and at the original SR) after the outboard.
This will make a realtime capture of split audio in it's own folder.
I'll then use soundblade's SRC to down sample those two wav's to 44.1k. (Off line)
Then I'll export it, bringing down the EDL desk a further 0.1db before the dither to 16. (Off line also)

I've never had an issue with regards to sound and I never get overs doing that. (on that setup anyway)

Cheers
Ade