ADT-Automatic Double Tracker Simulator

[Mark Hammer]

I feel like I'm at a kids' birthday party, watching "pin the tail on the donkey" and everyone is sticking the tail on the donkey's face or back.  Close but no cigar.

Whether it is tape, digital, Leslie, analog or rubber bands and masking tape makes little difference.  The key consideration is what makes something *sound* like two distinct individuals doing the same thing.  Chorus pedals have historically been an attempt to simulate the sound of....well, a chorus...several individuals singing/playing the same material in synchrony, or at least attempted synchrony.  One copy of the signal "takes the lead" and a second copy falls behind and catches up in periodic fashion.  You can achieve that by slowing down an analog sample via a bbd, a digital sample, or by twiddling the motor speed control on the tape deck.  Hell, if you wanted to, you could start two turntables off with the same track and twiddle the turntable speed control in the corner and do it too.

Is that enough?  Not really.  One needs to keep in mind what the human ear is able to derive and what its attention is drawn to as the stagger between multiple versions changes.  If the stagger is very close, what is noticed is the notches/cancellations created by two versions of the same content closely spaced in time.  That's flanging.  If the stagger is slightly larger/longer, perception of the notches declines and perception of pitch deviations tends to dominate, as well as an audible asynchrony.  That's chorus. 

What people generally refer to a "automatic double-tracking" is a similar kind of stagger but somewhat longer, probably more in the 20msec+ range.  How is this different, and how is it different than slapback?  Easy part first.  With slapback, the amount of stagger never changes, whereas with double-tracking it does.  The second voice is always discernible as a repeat with slapback.  Now the harder part.  With ADT, the stagger will occasionally (more in a second) get shorter such that it doesn't *always* sound like a discernible repeat.  But will the stagger be, or more importantly should it be, periodic like an LFO-based chorus?  I think the answer can be found by imagining yourself as a singer in a studio trying to repeat on playback what you just sang and recorded.  You listen through headphones (put on your best pensive look with hands clasping the headphones, kiddies, and crowd that spit barrier in front of the Neumann U47) and try to remember what you just did to the best of your ability.  How accurate will your pacing be?  In other words, as you try to recreate the parsing of words, the emphasis, the periods of silence, etc, what will your margin of error be?  Likely you will occasionally sing the same thing at almost the same time as the first track, but more often than not, you will be in the vicinity of 25-50msec off, simply because of reaction time, and relative accuracy in preparing movements/actions.  The end result will sound like two individuals because of that asynchrony, but here's the big magic difference in my view: the pitch deviation will be negligible because the stagger will be a function of the duration of silences or sustain of a note, NOT because of a graduated catching up and slowing down as in the case of a chorus.

It IS possible to get close to mimicking this with a sampled-delay (analog or digital).  I have one of those old blue rackmount MXR digital delays, and the LFO modulation can be applied to any of its 8 delay ranges.  Fortunately, because digital delays of that era had limited delay-modulation (the MXR swept over a maximum 4:1 range), it was easy with that unit to set it for, say, sweep between 30 and 40msec delay.  The sound is one I like to describe as being a bit like Pat Metheny, which is chorus-like but has a different ethereal quality that comes from the easily discernible stagger between wet and dry.  I haven't heard the "bounce" function on the old Eventide delays that Steve Giles is fond of, but I imagine that might introduce an element of realism too, and is possibly what distinguishes tape-based ADT from totally electronic attempts.

Ideally, though, an ADT would have only the faintest pitch modulation, and would probably use time compression to vary the duration of the gaps between notes.  That is probably not the sort of thing you can do in real time, but is consigned to the realm of post-production.  Choruses, of course, are kind of stuck when it comes to doing such things, because they have to always speed up the sample output of the audio to catch up and slow it down to lag.  While this will shorten and lengthen the gaps between notes, it will also distort the pitch as well.

For the time being, though, and for the adventurous, simply double the clock capacitor on your chorus pedal, to shift the range over.  You can easily tack a second parallel cap on the copper side of the PCB for a brief experiment.  You will notice two things immediately.  That whining.  Please, God, stop that whining!!  It's driving me INSANE!!!   The other thing you'll notice is that your depth control will need to be turned way down and may not even have the degree of sensitivity needed to dial in just a touch of modulation.  These two phenomena result from a) the clock frequency being much lower than the lowpass filtering can filter out, so much more audible, and b) the perceptibility of pitch changes at different delay ranges  (flanging changes pitch too, but you don't really notice it because at that delay range it is minimal; longer delays = more pitch deviation).

[alteredsounds]

Have a look for my thread about the '70's Resounder stereo ADT / Flanger I posted a couple of weeks ago.  Very nice sounding.  Might be 'clonable'.

[LyleCaldwell]

I feel like I'm at a kids' birthday party, watching "pin the tail on the donkey" and everyone is sticking the tail on the donkey's face or back.  Close but no cigar.

Mark,

I think that's not quite fair.  Some of us were talking about emulating the EMI ADT, not a box that would just sound like a doubled vocal.

[Ge_Whiz]

On a more practical note, I recall that the Danelectro 'Reel Echo' pedal has a 'wow and flutter' emulation which might get you close to, er, something.

[Mark Hammer]

I feel like I'm at a kids' birthday party, watching "pin the tail on the donkey" and everyone is sticking the tail on the donkey's face or back.  Close but no cigar.

Mark,

I think that's not quite fair.  Some of us were talking about emulating the EMI ADT, not a box that would just sound like a doubled vocal.

Well, you're right.  It was little more dismissive than was warranted.  Apologies for the appearance of harshness.  On the other hand, even the EMI-ADT is attempting to mimic something, so for me one has to start by asking "How do humans perceive the phenomenon we're trying to emulate?", since it is a psychological phenomenon we are trying to emulate (or emulate an emulation of).  Once you click into the auditory cues that people use to infer that it's two individuals playing/singing the same material, it becomes a lot easier to figure out what sort of technological requirements could achieve that, or come closer to doing so. 

Your suggestions re: the various digital delay lines that have modulation capabilities are bang on, and line up with my experience using the MXR unit.  In theory, anyways, it should be possible for any delay line that can apply subtle modulation to that delay zone between 20 and 50msec or so to get reasonably close approximations.  If the modulation can be made less periodic, via electronically-derived random fluctuations, or via hand/foot control, so much the better.

[LyleCaldwell]

I've had good luck using the Eclipse for ADT using the random LFO waveform.  Still periodic, but as it doesn't have any rhythmic correlation to the music played, the ear doesn't perceive a periodic pattern.

But that's an $1800 device.  And the subleties that are apparent on a recording would be lost on stage. I would bet most people would find a Deluxe Memory man set to no feedback with a fast delay time in vibrato mode to be very convincing (particularly if you use "my" mod and replace the 100K input resistor with a 1M pot and keep it in the 500K-1M range so pickups don't get so loaded down).

[Dirk_Hendrik]

One device with the nature that Lyle describes comes to mind. The Carlsbro ADT, which is, when set in the ADT setting like a chorus which can be set (rate at almost 0) to a chorus without the (percievable) modulation. Especially for UK forumites (as they're sold most in the UK) get one:

[vanessa]

I have found (for vocals anyway) the easy way to get a perfect double tracked vocal without having to do 50 takes (in fact I can normally get them on 1 or 2 takes) is to do it this way:

(BTW thank God for DAW's!)

I have the vocalist get the perfect take for the first track (obviously),

I then have them record a second "doubling" take (hold down the laughter just for a moment).

I usually find the big problem with double tracking is at the very beginning of or at the very end of a phrase. Usually the problem is lining up the very first uttering of the line that's being sung with the original perfect take track. Also at the end of a phrase, say when one's holding a note and overshoots or undershoots the phrase. If they overshoot at the end you can very easily fade out the double track with the original.

If they overshoot or undershoot at the beginning of a phrase I just copy and paste the very first uttering of the phrase from the first track to the very beginning of the second "doubled" track. Then I use a cross fade on the second track to smooth out the transition. And for just a few seconds of work I got instant ADT. What's cool is it has all the natural modulation of two people singing the same phrase, but with out all the headache of countless takes.

I'm sure this is done all the time in studios. But in case someone here is not familiar with it I thought I'd post it.

[SeanCostello]

The end result will sound like two individuals because of that asynchrony, but here's the big magic difference in my view: the pitch deviation will be negligible because the stagger will be a function of the duration of silences or sustain of a note, NOT because of a graduated catching up and slowing down as in the case of a chorus.

OK, try this experiment: Record yourself singing a single note. Then, double track this note. Try to get NO beating between the two notes.

Unless you are a freaking robot, this is impossible. The human voice has a certain amount of random jitter in it, and tends to drift all over the place. You will ALWAYS have beating between notes, regardless of the timing between individual notes or syllables.

For guitar, it is somewhat easier to have the same basic pitch between notes, if you match up the tuning correctly. However, unless you pick each note in a double tracked line in EXACTLY the same position with EXACTLY the same amount of force at the EXACT microsecond as the original recording, you will have phase differences between the notes, which is what you have with mild amounts of chorus anyway.  Plus, guitar strings almost always have a little bit of attack-dependent pitch bend, due to tension nonlinearities, which may be too subtle to hear on their own, but will cause beating effects when double tracked.

Sean Costello

[Mark Hammer]

I yield.

But how would one mimic that?  I don't know about you, but I've NEVER heard beats in any of the chorus devices I've used, and there's a lot of em.

[SeanCostello]

I yield.

But how would one mimic that?  I don't know about you, but I've NEVER heard beats in any of the chorus devices I've used, and there's a lot of em.

Beating = detuning. Sinusoidal driven choruses will have time-varying beat patterns. Triangle wave (Roland Dimension) and pitch shifter based choruses will have a steady number of beats per second for a given input frequency.

The easiest way to emulate "true" double tracking would be to simply emulate the overall time lag and detuning, using a longer delay line (20-40 msec) with a semi-random LFO, with a fairly low depth. Thiw would be close to the Beatles ADT.

If you want to emulate the phrasing differences, you could implement a transient detector, and use it to crossfade between two delay taps, with a random delay time set every time there is a new transient. This would generate the widely varying delay time Mark discusses, without the excessive pitch modulation that would result from trying to generate this variation with a single delay tap.

Sean Costello

[vanessa]

Or you could do what I have posted above and be done with it...

[LyleCaldwell]

Vanessa,
That is one of many ways to do it after the fact in the studio.  I took the thread as asking more about pedals/effects to do it in real time.

[MartyMart]

Vanessa, there's a new plugin ( I forget the name ) which will "pull in" double tracks
and harmony vocals to "match" the lead track !!
It copies the timing from the lead to the others, at "sample" level !!
Just thought you'd lke to know ( saves a LOT of time )

MM.

[rockgardenlove]

I've created a monster thread 


So, it seems to me at least, a bit of vibrato, with a bit of delay, and a bit of chorus should come somewhere close?  All together with an irregular LFO or what?  Set the LFO off sync from the music?  I see there's definetely no stompbox for this yet...

I'd say a very fast shallow vibrato, with ~20ms delay(maybe have the delay length hooked up to an LDR LFO?) and then a chorus...how fast for the chorus?  Mid speed?


This would be fun to put in one big box

[vanessa]

Vanessa, there's a new plugin ( I forget the name ) which will "pull in" double tracks
and harmony vocals to "match" the lead track !!
It copies the timing from the lead to the others, at "sample" level !!
Just thought you'd lke to know ( saves a LOT of time )

MM.

Marty! I've seen one from Waves that lets you run the plug-in on the original track letting you do very minor pitch shifting, modulation, and slight delay. To me it never really sounded like true double tracking. I would be interested in checking out the one you're talking about. The last time I fiddled with ADT plug-ins was about a year or two ago and at the time I did not find anything that sounded like you would want it on a professional release. I'm sure they've made improvements.

As far as live/real time goes... There's a few rack gear units (and I'm sure pedals out there) that come close if not nail it for live use. It's when you got the 'cans' on is when you really can tell (on a recording it needs to be on the money). The Abby Road ADT does not sound like it was a 'real time' type machine. In a shoot out I'm sure I would be finished putting the finishing touches on my tracks (using my method) long before the valves had warmed up on their reel-to-reels. As far as that 'tape' or 'tube' vibe? Just record it to tape (tube preamps) and dump it into your DAW (Protools etc.). Even with dumping the tracks I would guess the process still would be faster with similar results.

[notchboy]

I feel like I'm at a kids' birthday party, watching "pin the tail on the donkey" and everyone is sticking the tail on the donkey's face or back.  Close but no cigar.

[vast amounts of text deleted]

And I feel like I'm at a birthday party where a kid walks up to the donkey, announces that everyone else got it wrong, and then pontificates for 45 minutes about "the tail definitely isn't here... or here... " then continues on about the physiology of the hindquarters of the donkey for another 45 minutes... 

Meanwhile, everyone else has taken their masks off and is having a great time eating birthday cake.   

[rockgardenlove]

I'm still plotting my effect, I'm choosing to skip over all this raffle.  No offense meant to anyone who participated in the spam


So, here's my proposal:

Vibrato-pretty fast, and shallow,
Delay-hook the delay length up to an LFO that varies between 10-20 ms or so.
Chorus-Speed undetermined..how deep too?  I'd say pretty slow though, and medium depth.  What are your opinions?

So this would be my proposed setup:

Code Select Expand

   in
     |
     |
   panner
   |    |
  dry  wet(effects)
   |    |
   \    /
     out
      |
      |



Now, questions I have that I'd appreciate some info on:

Should the effects be put in series or parallel?  And if in series, what order would be optimum?
General info on LFO speeds, etc.  Should there be one LFO for all?  Should certain effects share an LFO?  What do you guys think?


Thanks!

[Dave_B]

I feel like I'm at a kids' birthday party, watching "pin the tail on the donkey" and everyone is sticking the tail on the donkey's face or back.  Close but no cigar.
[vast amounts of text deleted]

And I feel like I'm at a birthday party where a kid walks up to the donkey, announces that everyone else got it wrong, and then pontificates for 45 minutes about "the tail definitely isn't here... or here... " then continues on about the physiology of the hindquarters of the donkey for another 45 minutes... 

Not cool.

[rockgardenlove]

So erm, sorry to intrude, but could someone pay the slightest bit of attention to my question and stop bickering?  Please?  Don't mean to seem demanding, its just that the meaningful posts get lost in the flood of all this other stuff.

Thanks