Topic: Merging/Matricing 2 sources

[dorrcoq]

I have two sources of the ABB 10/04/08 show, raw .wav files @ 24/96.  The info is:

Two SD722 recorders were C-Linked as a tandem pair.  The SD722 running MG210 microphones was designated as the master, the ADK A51-TLs were run from the second deck as the slave.  All four channels were driven by the master word clock to yield 4 completely synchronous channels.

I would like to do a matrix of the two sources.  I have Audacity and Cool Edit Pro 2.  Can anyone give me step by step isntructions, please? 

Thanks

Dennis

[Roving Sign]

In a nutshell...for Audacity

Project > Import Audio (x2...one for each track set)

Click the "Time Shift Tool" and use it to align the second track set with the first. (click and drag)

You can adjust the gain in 3db increments on the sliders...these are good for playing with the blend (if you want finer control...use the "Amplify" tool from the Effects menu)

Save and Export...

Pay attention to what happens to your levels when the sources start to sync up - since you're in 24 bit you probably wont have any overs...but watch for red lights on the Audacity meters...peaks can combine and become even stronger peaks....

As far as something to sync to - I find music with some cymbal shimmer works well, listen for the bell part of the cymbal. Cymbals seem the most obviously affected by small timing moves.

[dorrcoq]

Thanks!  I'll give it a shot.  The times are already synched exactly so that shouldn't be a problem.  I hope.

[Roving Sign]

Thanks!  I'll give it a shot.  The times are already synched exactly so that shouldn't be a problem.  I hope.

Sounds as if they are clock synced and drift free for sure - but I'll leave it to the 722 owners from here - I dont know the exact details of the C-Link setup. Does it allow for two records to be started and stopped at the same time?

Sounds like your mic sets were pretty close together...so you are pretty close in time...

If you were at opposite ends of the room, you would probably need to time align, even if C-link allows for controlling mulitple recorders...

[page]

Sounds like your mic sets were pretty close together...so you are pretty close in time...

If you were at opposite ends of the room, you would probably need to time align, even if C-link allows for controlling mulitple recorders...

correct, you are *pretty close* in time, still should check and do fine adjustments where needed. Just a little off can create unintended effects. The two things to look for is clock adjustment (which it sounds like is not a problem here), and then delay adjustment which someone should *always* check. Depending on how far appart the mics are, very little if any may be needed, but you should still look and listen.

Also, you can control the slide volume adjustment in 1db increments, and even finer then that if you double click on it (unless its different in windows).

[RobertNC]

The Gefells were run on the top of the stand, DIN config.  The ADKs were run right underneath as M/S. 

I tried a couple of merges but did not really like any of them better than the Gefell source alone.  But I did not need to make any adjustments based on any kind of "syncing" consideration.

The file splits varied across the two decks as each one controls it's own file system, so you do have to join each set into a single wav before you do anything.

[easy jim]

The Gefells were run on the top of the stand, DIN config.  The ADKs were run right underneath as M/S. 

I tried a couple of merges but did not really like any of them better than the Gefell source alone.  But I did not need to make any adjustments based on any kind of "syncing" consideration.

The file splits varied across the two decks as each one controls it's own file system, so you do have to join each set into a single wav before you do anything.

Mix the m/s raw source to stereo first, as an independent source, before trying to blend it with the Gefell source.  Aside from lining them up on the time ruler, you should not need to correct any drift over the course of the recording since they were already linked/synced.  Listen back in real time and make lots of minor adjustments in the relative levels of the two stereo sources until you either find a mix you like, or decide that it's better to keep them as two discreet sources.

[dorrcoq]

...or decide that it's better to keep them as two discreet sources.

That might be the best idea of all!   I just thought that since Robert sent me both sources, and they were time synched, it migh be a good opportunity to try to do a matrix.  Maybe not.

[easy jim]

^ I think m/s + a near-coincident pair has the potential for making a good 4 mic mix...I have done a few of those that I really liked.  It's mostly a matter of fine-tuning to get the relative levels of the source correct in the mix and getting it to sound 'right' to you.

IMO, the 4mic will probably sound better if you decode the m/s to a more mono-ish vs. stereo-ish mix.  Let the near-coincident pair define the soundstage and the m/s (and LD) pair give it the oomph and fill in the center...

[RobertNC]

^ I think m/s + a near-coincident pair has the potential for making a good 4 mic mix...I have done a few of those that I really liked.  It's mostly a matter of fine-tuning to get the relative levels of the source correct in the mix and getting it to sound 'right' to you.

IMO, the 4mic will probably sound better if you decode the m/s to a more mono-ish vs. stereo-ish mix.  Let the near-coincident pair define the soundstage and the m/s (and LD) pair give it the oomph and fill in the center...

That is a thought.  I went the other way and decoded the M/S first with a lot of the figure 8 then tried various levels of merges.  I think we were a little too far back to get a really good M/S, but I had the whole weekend run of shows to track out and ended up not spending much time with it.  I am on old laptop that has such limited hard drive space, I have to dump the masters from one show to start on another.  I need to upgrade but may just get an external HDD.

Been meaning to go back to this show and play with it some more, but have not done it yet.

One thing about this show - I was a little disappointed with the ABB sound.  It seems like Greg has everyone else turned way down.   I mean three percussionists on stage and there is barely any rhythm section in the mix.    It wasn't the venue - Phil and Molo came out and thundered.

Might get a little more low end from the card channel of the M/S pair, the Gefell hypers are pretty bright and the sound at the show was pretty, well, bland.

[RobertNC]

Aside from lining them up on the time ruler, you should not need to correct any drift over the course of the recording since they were already linked/synce

You don't have to worry about lining up the time once each source has been joined to a single wav.  I don't know the details but when you press record on the master there is some kind of handshake between the 722s, and the master and the slave exchange enough info about whatever is in the buffers to start writing the files on both decks on the same initial sample.  So while as has been mentioned you could be sampling different signals based on sound arriving at the various microphones, the sampling process itself is synchronous from start to finish.

(Or not.  In which case you get bad data.  We tried the same thing at a Ratdog show in Nov and had a cable problem or something.  We always put the best mics on the master in event of problems, and at this show the master deck ran fine but the slave deck had numerous dropouts).

[RobertNC]

OK here is what I using CEP.

First of all, I suggest you join each source, then they will be synched.  Then cut out a fairly short section that contains something representative - vocals and jams fom one of the louder regions from both sources.  Otherwise this will take too long to process to really be able to compare much. 

Also, unless the L/R balance of an individual source seems off (and if I remember, both of these were reasonably close) I don't worry so much about stuff like normalizing each source to some value.  Instead I just mix them as is and listen and make sure I keep track of it all so I can reproduce everything I did later.  Your call on how you like to proceed.

So...

Step-By-Step: (And keep in mind I am no CEP whiz, so if someone can improve on this, please do).

Step 1)  Decode the Mid-Side ADK source first.  (Obviously this requires the most work).

The left channel is the Mid Cardioid.  The Right channel is the side Figure 8.

Mid + Side = new Left Channel
Mid - Side = new Right Channel

To Decode the M/S source into L/R in CEP Select Effects > Channel Mixer.  There is a preset for Mid-Side to L/R.  Select that option.

CEP M/S decoding is a little confusing, because they don't give things in "ratios".  When you first select the preset you will see 4 sliders.  The new Left will default to 100L, 100R.  The new Right will default to 100L, -100R.

The CEP documentation on this sucks, but from what I can piece together, I *think* the following is true, not absolutely positive:

This default is equivalent to a 50:50 Mid-Side decode.  Decoding does not change the total signal (although I think this might be based on keeping something else constant like RMS so you may see some changes in the final dB levels.)

According to CEP help using channel mixer functions should not add gain, but I usually see a bump of maybe 2 or so dB in the final L/R channels.   For 24 bit the final decode will still have lotsa head room to work with later.

If you set the new Left channel to 100L, 0R, new Right Channel to 100L, 0R, the result will be a "mono", or more correctly, 100% mid (cardioid only) L/R.

If you set the new Left channel to 0L, 100R, and the new Right channel to 0L, -100R the result will be a 100% side (Figure 8 only) L/R.

Anything else is somewhere between.

To increase the Mid:Side, I would suggest you leave both left channels at 100L for Left, and start decreasing the Right from between 0-100R for the new Left channel and -100-0R for the new Right channel.

Ditto for decreasing Mid:Side.  Leave right channels at 100R, and -100R for new Left and new Right and start decreasing the Left from 0-100L for both new Left and new Right.

In the first case you are actually leaving the Mid at "full", but decreasing the Side, which results in higher Mid:Side ratio.  In the second case you are leaving the Side at "full" and decreasing the Mid which results in a lower Mid:Side ratio.

(Again this is a little confusing because CEP does not use the typical "ratio".)

One thing for sure:  even though CEP will let you do it, do not go outside these bounds:

Left for both new Left and Right: 0-100L.

If you go below 0, it does not "reduce" the Mid anymore, it starts inverting it's polarity.

Same is true for Right for new Left and Right except here you have plus and minus so:

New Left:  Keep new Right  between 0-100R, or you are inverting.
New Right:  Keep new Right between -100-0R, or you are inverting.

This will make a lot more sense once you start working with it.  Play around starting at extremes and working inward.  Once you decide on a ratio, you may want to tweak the final L/R a little if needed.  I suggest you don't bump it up because the next step is "additive".

Step 2)  Start merging the two sources.

Again it's better to be working with a short section because you may want to go back and try a different M/S ratio after you try merging.

In CEP, open the Gefell source wav section that matches the ADK section.  Of course since it is a normal L/R stereo source you don't need to decode.  If you feel the need again, tweak L/R if you want.  You should now have both sources open.

Percentages will not help you here.  As far as I know Multi-Track view can only be used in dB format.  Again not a CEP whiz.  If you decide that you want to bump gain first on both sources, you'll need to do this subtractively.  I am assuming original 24bit sources with low intrinsic gain.

In menu select View, Multitrack view.  Empty multi tracks appear.  Put the mouse in the Track 1 window.  Be careful not to click - it will move the cursor.  With the mouse inside Track 1, right click, then a drop down appears, select "Insert" and you will see your two open wavs. Select the ADK source and it will be copied into Track 1.  Repeat for Track 2 using the Gefell source.

Now here is the tricky part.  As far as I know there is no way to start mixing with specified ratio - I only know how to mix down all tracks.  For the typical way I usually run the 722 you probably have maybe 8dB headroom to work with.  You can control the final mix ratio by setting the volume on each individual track.

This Mid-Side source is not very good, definitely not a primary source.  So you will almost for sure want more of the Gefell.  Again this process is additive.

Beside Track 1 you will see a box that says V0.  Right click on that.  For  starters bump that up to 2.5 dB, assuming that is the ADK source.

Beside Track 2 do the same except bump that to maybe 5 dB, assuming that is the Gefell source.

Now put the cursor in Track 3 (empty) and right click.  This time select "Mix Down To Track (Bounce) and then "All Waves".

The result in Track 3 will be Track 1 at whatever gain + Track 2 at whatever gain + 7.5dB additional gain added to the additive gains already present, but with 2.5 dB added to the ADK source and 5dB added to the Gefell source. Gives you roughly a 75% Gefell/25% ADK mix.  Listen, save and then purge Track 3 (or not), repeat.

If you want to try mixes without saving and clearing Track 3 every time then make sure when you mix all waves into, e.g. Track 4, you also set the volume in Track 3 down to -infinity.  This will allow you to keep the Track 3 mix without anything being added from it to your next attempt.

Sorry this is not very exact.  Only attempted the 4 mic mixes a few times, and not in good venues. 

And as far as post - I am a taper, not a producer.

If anybody knows a better way, please post.

[dorrcoq]

YIKES!

[RobertNC]

It's not that bad once you actually do it a couple of times.

[easy jim]

Decoding does not change the total signal (although I think this might be based on keeping something else constant like RMS so you may see some changes in the final dB levels.)

Decoding raw m/s results in some summing in the stereo mixdown, up to +3dB or so from the relative values of the raw sources in my experience with Digital Performer/AudioDesk - and it seems that should hold in general.
___________________________

Percentages will not help you here.

Agreed.  There's no need to approach a 4mic mix IMO from the strict perspective of mixing x percent of one stereo source with y percent of a second stereo source.  Once you're mixing additional source(s) with a main stereo pair, you might as well get creative if it can help you produce a better result than the raw sources.

If the m/s source does not sound very good, I'd first try remixing it to a more mono-ish mix since you were further back in the room.  Hopefully, the low end of the m/s source is nice and coherent and only the higher frequencies suffered from being too far back.  Considering that the Gefell source sounds brighter, I'd try mixing it with the low-end of the m/s source (filtering some of the high frequencies with a shelf filter).  Assuming that the low end of the m/s source is more coherent, you can use more of it (scooping some of the highs) in a mix to emphasize that characteristic of the source in the 4mic mix to give the final mix more of the low from the m/s source without adding too much of the less-coherent m/s source highs that would detract and 'smear' the more-coherent highs from the Gefell source.

[RobertNC]

Decoding does not change the total signal (although I think this might be based on keeping something else constant like RMS so you may see some changes in the final dB levels.)

Decoding raw m/s results in some summing in the stereo mixdown, up to +3dB or so from the relative values of the raw sources in my experience with Digital Performer/AudioDesk - and it seems that should hold in general.
___________________________

Yeah, I  have corrected it. According to the CEP help channel mixing is not additive, but I usually see a couple of dB or so increase as well.  I've never liked the approach CEP uses to M/S as opposed to the usual "ratio" concept, but I don't run it enough to really work it all out or need a better DAW.