Topic: Matrix in post best practice workflow

[tim in jersey]

n00b to the matrix in post thing...

Scenario:

Deck 1 is a SBD feed, Deck 2 is my aud recording. Neither is jammed to timecode. I mix/correct for drift in post using Vegas.

I run levels on each very conservatively. About -9 to -6 dbfs, peak.

Should I normalize/amplify each file prior to bringing it in to the DAW or just mix and then amplify?

[justink]

i amplify both sources to about -5 or -4 db (for a 50/50 mix).  then align.  then mix/render (the joining will amplify some waves).

there's a great thread about this using audacity, pretty easy.

[page]

I use the following:

1) Align via preferred method (stretch in Vegas/Reaper or time-align in Audacity)
2) Determine what I like about either source (sbd has less reverberation in the 1-3khz range, aud has a warm gooey bass, etc), and EQ to bring out what I like about both, maybe a net change of 2db tops, just enough to nudge it in a direction. Don't overdo it.
3) Make sure the peaks in either are at least -6 or lower.
3.5) double check alignment.
4) Render to 2ch. (Assuming a render is required to apply editing to both sources equally)
5) EQ/process anything else.
6) Amplify the remaining db and do any limiting and dither.

I used to keep the peaks between -3 and -6 which mathematically should bring your render to just under 0, but I kept having the occasional overage and having to redo the render so I figured the additional few DB of amping in Ozone or whatever is clean enough I'm not worried about it. Make sure to double check your alignment, I prefer to check mine now over the course of a couple of listening sessions and in both loud and quiet sections throughout the recording since non-clocked sources can drift at different speeds (ever so slightly which is a pisser but usually not much of a problem). I don't think I've forgotten anything....

[justink]

I use the following:

1) Align via preferred method (stretch in Vegas/Reaper or time-align in Audacity)
2) Determine what I like about either source (sbd has less reverberation in the 1-3khz range, aud has a warm gooey bass, etc), and EQ to bring out what I like about both, maybe a net change of 2db tops, just enough to nudge it in a direction. Don't overdo it.
3) Make sure the peaks in either are at least -6 or lower.
3.5) double check alignment.
4) Render to 2ch. (Assuming a render is required to apply editing to both sources equally)
5) EQ/process anything else.
6) Amplify the remaining db and do any limiting and dither.

I used to keep the peaks between -3 and -6 which mathematically should bring your render to just under 0, but I kept having the occasional overage and having to redo the render so I figured the additional few DB of amping in Ozone or whatever is clean enough I'm not worried about it. Make sure to double check your alignment, I prefer to check mine now over the course of a couple of listening sessions and in both loud and quiet sections throughout the recording since non-clocked sources can drift at different speeds (ever so slightly which is a pisser but usually not much of a problem). I don't think I've forgotten anything....

he said it best.

[runonce]

n00b to the matrix in post thing...

Scenario:

Deck 1 is a SBD feed, Deck 2 is my aud recording. Neither is jammed to timecode. I mix/correct for drift in post using Vegas.

I run levels on each very conservatively. About -9 to -6 dbfs, peak.

Should I normalize/amplify each file prior to bringing it in to the DAW or just mix and then amplify?

If they are 16 bit sources - you might want to skip normalization until the end (as recommened by page in the last post) - You're going to need about -6db of extra headroom to handle the higher peaks that result when you combine sources.

[tim in jersey]

That answers my questions. Thank you, gentlemen.

[Walstib62]

I could be totally wrong on this, and if so, please someone let me know. But, I have always imported the tracks and normalized them first. The reason being that normalizing removes any possible dc offset. That way, all tracks are nulled to 0 offset prior to any other processing. To make things easier level-wise, I normalize to -6bB. Then, when the tracks are combined, the result is always a little below 0 without going over.

[justink]

I could be totally wrong on this, and if so, please someone let me know. But, I have always imported the tracks and normalized them first. The reason being that normalizing removes any possible dc offset. That way, all tracks are nulled to 0 offset prior to any other processing. To make things easier level-wise, I normalize to -6bB. Then, when the tracks are combined, the result is always a little below 0 without going over.

nothing wrong with that.

[page]

I could be totally wrong on this, and if so, please someone let me know. But, I have always imported the tracks and normalized them first. The reason being that normalizing removes any possible dc offset.

I just checked in Audacity and there is an option to remove the offset when you do it (news to me). I don't recall there being one in reaper, and I know in RX it's a separate function. In Ozone, they have it chained to occur at the end before dither or the output gain structure.

but you bring up a good point; do you strip it before you do anything or at the end? All of my current gear produces so little offset naturally that I haven't had much reason to do it any earlier in the chain, but from a best practices standpoint, it's a valid question.

[morst]

I could be totally wrong on this, and if so, please someone let me know. But, I have always imported the tracks and normalized them first. The reason being that normalizing removes any possible dc offset. That way, all tracks are nulled to 0 offset prior to any other processing.

I think you might be totally wrong. Normalization just brings up overall output levels, DC Offset Removal is a separate function. Unless you are using some offset-correcting normalizer plugin?

You'll get WAY better results mixing 24-bit signals if you can master that way...

[kubacheck]

I could be totally wrong on this, and if so, please someone let me know. But, I have always imported the tracks and normalized them first. The reason being that normalizing removes any possible dc offset. That way, all tracks are nulled to 0 offset prior to any other processing.

I think you might be totally wrong. Normalization just brings up overall output levels, DC Offset Removal is a separate function. Unless you are using some offset-correcting normalizer plugin?

You'll get WAY better results mixing 24-bit signals if you can master that way...

the one time I tried to do a matrix in Vegas, it wouldn't let me render the matrix of the two 24 bit sources as a resultant 24 bit output file, it automatically dithered it down to 16/44.1.... if there was a way to keep it at 24/48, I couldn't figure it out....

[morst]

it automatically dithered it down to 16/44.1.... if there was a way to keep it at 24/48, I couldn't figure it out....

If that's really true, I would advise you to find something other than Vegas to do your work. No sense in wrecking your bits like it's the only option!!?

[kubacheck]

it automatically dithered it down to 16/44.1.... if there was a way to keep it at 24/48, I couldn't figure it out....

If that's really true, I would advise you to find something other than Vegas to do your work. No sense in wrecking your bits like it's the only option!!?

its been a couple years, but as I recall, the pull down menus only went up to 16 and 44.1, I could've rendered at lower bit rates, but 16/44.1 were the highest quality values I could select..... perhaps that was specific to the version of Vegas I was using, I think it was Vegas 8.....

[beatkilla]

Thats definetely not true for Sony Vegas. You can work with and render up to 24bit 192khz.      I just checked and you must have the lite version of vegas and that one goes no higher than 16  48.

[kubacheck]

Thats definetely not true for Sony Vegas. You can work with and render up to 24bit 192khz.

well, like I said, if there was a way ( and I looked and looked), I could not figure it out..... the details are not fresh since it's been a couple years, but I recall I could not find any options anywhere that allowed me to render the output audio file at 24/48..... as I mentioned, I was using Vegas 8, if that means anything....