Topic: Help with making a matrix...

[JSouth]

Hi, first I am new, I was pointed here by someone on phishhook.

Secondly I am trying to make a matrix of a really good show. I have a nice aud and a good board of the show.

I am using cool edit. Is there another program I should be using?

The sbd disc and the aud's tracking does not match up.
I have tried lining up one track and then just putting the rest together since one spot is synced up the rest should just fall into place, correct?



Thanks for any and all help, Joey.

[jhirte]

I've used CEP to sync aud/sbd tapes.
Most recently was Caustic Resin 08/28/03. Board was mono though. bah.
anyway.
what I do is just find a point right at the beginning to reference, like a snare hit, a cough whatever.. and thenfind the time point right where it starts/ends. get both aud/sbd to have that same start point.
then I'll dump in multitrack and play both together to see how the sync is.

Theoretically on DAT they should be the same speed, etc...
but on the caustic resin show I taped, one tape ran a hair fast. literally like THOUSANDTHS of a sec, which was enought o throw it off.
board > D8, naks > DA-P1, and DA20MKII for playback > PC.
I got them sync'd fine, but was a pain.

Some shows you'll do will be perfectly fine though... now, get some crazy old analog tape and mix that with another.. that'd be fun!!! haha

-Jim

[nickgregory]

I have done quite a few matrices (sp?) and here is how I would do it.  Use CD Wave to cut up the SBD cut into tracks, but make each track start with something significant, a cough, snare hit..whatever.  The use the multitrack view on Cool Edit, open up the AUD source as one wav file and mix it track by track in the multi track view.  By doing this, you will fix the couple of second variations that you will likely have.

Nick

[HanKDanK]

sometimes you will have to break down the tracks very small... especially if it is late in the show and the room is hotter, more smoke, more moisture, etc... and the sound travels to the mics differently.  Cool Edit Pro is the way to do it, just pick a few spots and zoom in and line it up, real easy, but can be time consuming.  Sounds f-ing great if you pull it off right though  

-Hank

[joel]

Wassup Joey?  Good to see you on here!  I haven't had any luck doing any sbd/aud mixes after the fact.  My latest recorded in real time from over there in your neck of the woods in Carborro, which came out great (SBD + Onstage Neuman TLM 103's > Soundcraft Spirit M8 > Oade Modified UA-5 > JB3 Optical Line-In .Wav @ 44.1khz
) is up on the etree archives here:

Perpetual Groove: September 4th, 2003

Nice logo Nick!  Y'all hittin the Harmonized Showcase at the Lincoln in Raleigh in a couple of weeks with Ulu and the Motet on fri night and Lotus, PGroove, and Garaj Mahal on Sat night?  Speakin of GM, how did those Smilefest vids turn out Joey?

Joel (Jhudson844)

[nickgregory]

unfortunately I am going to miss the harmonized showcase...going to be out of town til Moe on Sunday....

[dklein]

Yeah - it sucks but without synching the time clocks (like on pro gear), different sources never seem to match.

Depending on the amount of drift, cutting it up can be really tough.  It's not always easy to find a point to align and you'll find that there is some fluctuation depending on which sound you're matching.  I have used the stretch feature in both CEP and wavelab to make one the same duration as the other.  You want to preserve the pitch and get them within a few milliseconds of each other.  I couldn't get CEP to stretch without introducing some nastiness to the sound (and I tried lots of settings) but my most recent effort with Wavelab 4.0c seems to sound good.  If you're going to play with it, be sure to listen to both loud and quiet passages before mixing it in.

You can use CEP in multitrack mode to line up a point in the beginning.  
Then go to the very end and determine how much shorter/longer the 2nd source is.  I'd recommend using the stretch on the source that you'll be mixing in at a lower level.  Add/subtract that to the existing length and start playing with values in the stretch menu to get to the desired length.  I can tell you that putting in precise time values does not work.   Both CEP and Wavelab seem to only be capable of working in discrete steps, as defined by changes in the ratio box.  I learned this through many failed attempts at specific values until I began to recognize it's either A or B but nowhere in between.  If you find you can't get an acceptable value, you can increase the precision by splitting the file first (like into disc1 and 2).  You don't have to do the stretch to get the software to display the time you'll end up with - just play with the ratio settings.

Wavelab - go process, time stretch, check high quality and preserve pitch
CEP - as I mentioned, I just couldn't find a setting that sounded good throughout but give it a shot

Then multitrack with your stretched version and try and line up your peaks - you really need to zoom in and get them within about 10 ms throughout.  You'll find that your mic recording has peaks that are not nearly as well defined as the board and this task is a challenge.  One feature that I love about the new CEP 2.1 is the ability to slide the track while you're listening to it.  You can listen for yourself what kind of 'error' range you have before the sound really starts to screw up.  That's where I get the 10 ms above, though often the mics are so 'sloppy' and full of reflected sound that it can be more forgiving.  I thought it was more until I had CEP 2.1 and could listen to the effect live - listen to vocals and cymbals to hear what happens.

This exercise made me kind of down on the live matrix unless you're onstage or very close - once you hear the blur from 25ms you'll get it.  And I also discovered that the PA itself is often delayed a bit so that it synchs up with the natural sound (because the PA usually sits in front of the band).  One of my regular places puts a 7ms delay on the PA.  So if I tape 30 feet back, I've got a 37ms delay against the board feed.  Too much me thinks.

Hope that helps you out.  It takes a lot of hours to do a good mix.

Here's the related thread on taping - never got too far though http://www.taperssection.com/yabbse/index.php?board=23;action=display;threadid=6516;start=msg72858#msg72858

[JSouth]

Hey. Thanks for all the advice!

Joel the vids are ok nothing to write home about though.

Actually what I am trying to matrix is the KDTU from smilefest. I know the taper and maybe I can get a source that is not tracked out, but I doubt he still has the original copies due to taping on a laptop.

I just want to make the SBD sound a little more alive!

[Kwonfidelity]

I've done dozens of post matrices before and be careful when people tell you to match the beginning of the show then look at the end and you can calculate your drift.  You have to keep in mind that drift isn't always constant in a dat player, so if you match the front and then look at the end of the show (say it's a few msecs off) you also have to look at the what happened in the middle.  I've had the best of pro dat players give me drift in the middle of the show, so if you're using stretch functions you will still get reverb and phasing in that area where the drift has fluctuated throughout the whole show.  I'm not saying that the advise of beginning to end isn't valid, just make sure that you're not getting drift in the middle either.

I find that the easiest, most effective way to create a great matrix is by lining up the beginning of both sources, then track down the timeline every 3-5 minutes.  When you start to get drift, zoom into that area within 1 second and find a distinctive peak (snare, drum, etc).  Splice the source that is *longer* right before that peak/sound and shorten it to match the "tracking" source(the one that remains unspliced).  The chances are that drift is a slower dat, not a faster one - it's also audibly better to splice and compress a wav then it is splice and stretch one since you will end up with a void in the spliced source at that point.

Always try to avoid using stretch functions in software.  This  degrades the audio, no matter what kind of software you are using (even in PT6).

[dklein]

Thanks for the tips.  I haven't found the 'drifting in the middle' thing but I guess it may depend on the equipment.  One time I saw it but it was a result of a glitch in the transfer that had been edited by the guy that had the mic recording.  When I got another copy with the glitch, things were ok.

So how much drift do you think is acceptable before a cut is required to line them back up?

[JSouth]

I have finished but now have to go back and line up a track where there is only talking.  Sounds great so far. Thanks for the advice.

[dklein]

So how much drift do you think is acceptable before a cut is required to line them back up?

quoting myself again...The reason I ask is because I just did one that had 1.2 seconds of drift over 2 hours.  That's 10 ms per minute and I couldn't see myself cutting the file every 30 seconds.  

I was quite happy to find the wavelab stretch function sounded so much better than anything else I had tried.  After your post I went back and listened to the stretch and non-stretch and honestly can't distinguish the two.

btw, the 2 sources were a JB3 analog in and a laptop/UA-5 combo.  I don't suppose there's anyway to synch up or adjust this kind of equipment before recording... is there?

[Kwonfidelity]

How much drift?  Don't know.  I'm never that technical with anything  

I usually go strictly with what you can hear, therefore the playback is critical.  Once you start to hear the reverb then it's time for a splice, but you don't want to let it run too far b/c once you splice it right there the tones will change dramatically and will make it hard to listen to over time.

Sometimes you will end up cutting a show many, many times.  This is just a labor of love, or trash the project and live with the one source.   It all depends on how much time you have to spend on it (or how much someone is wanting/paying you to do it).

At best, you need timecode to lock two different sources together...

[dklein]

After doing a few of these, I've got one that I'm really happy with.  If anyone wants a listen, I just started a Bit Torrent http://tinyurl.com/rfgv

• The board was UA-5>laptop (24/44.1) and the mics were 16/44.1 (analog into the JB3).  
• I stretched the mics to match the board with Wavelab (temp set to 32 bit float) and saved the 32 bit float file.
• Mixed the 2 sources (also in 32 bit fp) using CEP 2.1 (I used some EQ to rolloff the bass on the mic source).
• Saved the mixdown in 32 bit fp and dithered to 16 with Wavelab.

It appears that my stretch is always the same value when I use this gear (which is convenient since calculating the stretch is a pain.  I can't do the editing thing because the drift between my sources is consisently 11ms/minute - meaning I'd have to do lots of cuts.  Also, I really think the Wavelab stretch sounds good and use it on the less dominant source.

So after playing for a while, I like Wavelab for stretch and dither, CEP for mixing and EQ.

If anyone gives it a listen - tell me what you think!

david

oh...if you don't know Calexico, think vacation in Mexico at an 'American' style bar  

[Kevin]

cant help you out but +T and welcome to the board