ESTIM DC conversion question (paging Kelvinator or LG)

[samsonjude1776]

So I have been enjoying DC conversions of triphase files as they often run better on y 312B but when looking back on the instruction in threads I noticed LG mentioned inverting the LT mono track for the second mix and Kelvinator said right. Can one of you clarify? And what difference does it make?

[Kelvinator]

Hi SamsonJude,Use LondonGent's instructions and invert the Left Track ( viewtopic.php?p=327182#p327182 ). Audacity has had a few updates (latest version 3.4.2) since these instructions were written, so you may have to search and poke around the menus to find the various track commands in the menus. Using LondonGent's instructions and Audacity Version 3.2.4, here's what to do:
1. Split the Stereo Track: Click on the Track drop down menu "V" arrow (not the Audacity top menu bar), and Select "Split Stereo to Mono".
2. Select Both Tracks: Hold the Shift key and click the Select button on each track. Mix and Render to New Track: From the top menu bar, click Tracks, Mix, Mix and Render to New Track.
3. Select the Left Track and Invert: The Left Track is the top mono track. Select it. From the top menu bar, click Effect, Special, Invert.
4. Select both tracks again, Mix and Render to new track - this will give you a second amplitude-based track in opposite time to the beats.
5. Combine the two new tracks at the bottom into one stereo track: Click on the Track menu bar, and select Make Stereo Track.
6. Use Effects > Amplify to remove any clipping: Modified in Audacity 3.4.2: From the top menu bar, select Effects, Volume and Compression, Amplify. Go with the default Amplification factor (usually a negative number).
7. Save the newly created dual channel file: Select the new dual channel file. From the top menu bar, select File, Export Audio, Current Selection (Make sure to rename the file).
I may have missed some steps, but that's the gist of it. Good luck. K

[zebbg69]

This is also called "Left/Right to Mid/Side" and there is an Audacity plugin that will do it in one operation. It is a standard conversion in audio mastering.

My Linux machine has the plugin, and my Mac does not. I am not sure if it was included in the Linux installation, because I don't remember doing anything special. You might be able to find the plugin by Googling if you don't have it.

The plugin probably divides each resulting track by sqrt(2), so that the operation can be repeated to toggle from L/R to M/S and back without doubling the level on every round trip. (Otherwise, repetition gives you [L,R] --> [L+R,L-R] --> [2L,2R] ...)

Karaoke plugins also add and subtract stereo tracks to eliminate vocals, though they tend to add some light filtering. If you have a karaoke plugin, it is worth experimenting. The result may be usable, especially if it lets you turn off the filtering.

[samsonjude1776]

Thanks for the information.

[LondonGent]

So I have been enjoying DC conversions of triphase files as they often run better on y 312B but when looking back on the instruction in threads I noticed LG mentioned inverting the LT mono track for the second mix and Kelvinator said right. Can one of you clarify? And what difference does it make?

IIRC it shouldn't make much difference which one you invert. 1 + -1 = 0 and -1 + 1 = 0.

The main thing is the conversion to mono. In triphase files the sensations are caused by the difference between the two channels. Converting the file to mono causes the two channels to be added together, turning the phase differences into amplitude (i.e. volume) differences instead.

The reason for inverting one channel is to make the resulting signal strongest on the beat, rather than weakest on the beat. If the resulting file is weak on the beats, you can skip the inverting step entirely to solve that problem but most triphase files need the inversion to be timed correctly.

[zebbg69]

So I have been enjoying DC conversions of triphase files as they often run better on y 312B but when looking back on the instruction in threads I noticed LG mentioned inverting the LT mono track for the second mix and Kelvinator said right. Can one of you clarify? And what difference does it make?

IIRC it shouldn't make much difference which one you invert. 1 + -1 = 0 and -1 + 1 = 0...

Subtracting (L-R) will feel similar to the Left triphase channel, and (R-L) will feel similar to the Right. Whichever one you generate, you can always just swap its leads on your body to invert it, and it's the same as if you had subtracted the opposite way in Audacity, so you can just pick one and not worry about it. This should practically never make a difference, since we are flipping a modulated carrier--the carrier itself will flip upside-down inside the overall shape, but the overall shape will be unaffected. It will still swell and shrink in the same exact spots, unless the modulator and the carrier get close together in frequency or do something else unusual.

I say "feels similar" because each channel also draws a cross current from the common, whose intensity depends on how close the L/R electrodes are to the common vs. each other. But if you compare DC to triphase with common electrode far from the L/R electrodes, then the DC's (L-R) and (R-L) are basically the same as triphase's L and R.

[LondonGent]

IIRC it shouldn't make much difference which one you invert. 1 + -1 = 0 and -1 + 1 = 0...

Subtracting (L-R) will feel similar to the Left triphase channel, and (R-L) will feel similar to the Right. Whichever one you generate, you can always just swap its leads on your body to invert it, and it's the same as if you had subtracted the opposite way in Audacity, so you can just pick one and not worry about it. This should practically never make a difference, since we are flipping a modulated carrier--the carrier itself will flip upside-down inside the overall shape, but the overall shape will be unaffected. It will still swell and shrink in the same exact spots, unless the modulator and the carrier get close together in frequency or do something else unusual.

I say "feels similar" because each channel also draws a cross current from the common, whose intensity depends on how close the L/R electrodes are to the common vs. each other. But if you compare DC to triphase with common electrode far from the L/R electrodes, then the DC's (L-R) and (R-L) are basically the same as triphase's L and R.

The vast majority of triphase files shared here don't have any real difference between the left and right channels. All the sensations come from phase differences between the channels, rather than the amplitude of the channels themselves.

When the two channels are in phase, the power at the common electrode is double the power at the other electrodes, for example L = +1, R = +1, C = -2. Current flows from the left to the common and the right to the common, it does not flow from from left to right since both are positive or negative at the same moment.

When the two channels are fully out of phase, the power at the common electrode is 0, for example L = +1, R = -1, C = 0. At this point, current flows from left to right since they have opposite voltages and does not flow to the common.

When we convert a triphase file to mono, the left and right channels are added together and these phase differences become amplitude (i.e. volume) differences instead. The mono file will have it's maximum volume when the two channels are in phase and zero volume when they are in anti-phase. If we invert one of the channels first, it will shrink and swell in the opposite spots since we are reversing the phase differences. Whether you want to do that or not depends on whether the signals are in phase on the beats or out of phase.

As for whether to invert left or invert right, 99% of the time it won't matter since the files don't use the left and right channels independently anyway. 1 + -1 and -1 + 1 both equal 0.

[zebbg69]

IIRC it shouldn't make much difference which one you invert. 1 + -1 = 0 and -1 + 1 = 0...

Subtracting (L-R) will feel similar to the Left triphase channel, and (R-L) will feel similar to the Right. Whichever one you generate, you can always just swap its leads on your body to invert it, and it's the same as if you had subtracted the opposite way in Audacity, so you can just pick one and not worry about it. This should practically never make a difference, since we are flipping a modulated carrier--the carrier itself will flip upside-down inside the overall shape, but the overall shape will be unaffected. It will still swell and shrink in the same exact spots, unless the modulator and the carrier get close together in frequency or do something else unusual.

I say "feels similar" because each channel also draws a cross current from the common, whose intensity depends on how close the L/R electrodes are to the common vs. each other. But if you compare DC to triphase with common electrode far from the L/R electrodes, then the DC's (L-R) and (R-L) are basically the same as triphase's L and R.

The vast majority of triphase files shared here don't have any real difference between the left and right channels. All the sensations come from phase differences between the channels, rather than the amplitude of the channels themselves.

Yes, but this is entirely how triphase works. The summing you do in Audacity mimics what the electricity does in your body. What you feel at the triphase common is literally (L+R), because both channels draw their current from the common. Okay, not quite literally, because they are each scaled by resistance based on placement on the body, so it's (aL+bR) where a and b are scaling due to resistance, but (L+R) captures the essence of it.

Typical triphase files are very different between L and R. The difference is not "only" phase, it *is* phase. Each moment in time produces a sum, which as you say can be up to double the swing of the whole wave or can cancel to zero. If you zoom way out, you can't see the difference, because you can't see the phase anymore.

The reason your ingenious DC conversion works is that it computes the actual triphase effect so we can feel it on each channel independently without relying on triphase physics to make it happen. (L+R) is the current on a triphase common, and (L-R) is the current on triphase L and -R (mostly, as I noted above). We're saying the same thing physically, only the interpretation is different. What you've computed *is* the triphase effect, not just a sort of min/max/mono thing that approximates it, and I would not say that a triphase L and R are not very different. They are vitally different.

[LondonGent]

Yes, but this is entirely how triphase works. The summing you do in Audacity mimics what the electricity does in your body. What you feel at the triphase common is literally (L+R), because both channels draw their current from the common. Okay, not quite literally, because they are each scaled by resistance based on placement on the body, so it's (aL+bR) where a and b are scaling due to resistance, but (L+R) captures the essence of it.

Typical triphase files are very different between L and R. The difference is not "only" phase, it *is* phase. Each moment in time produces a sum, which as you say can be up to double the swing of the whole wave or can cancel to zero. If you zoom way out, you can't see the difference, because you can't see the phase anymore.

The reason your ingenious DC conversion works is that it computes the actual triphase effect so we can feel it on each channel independently without relying on triphase physics to make it happen. (L+R) is the current on a triphase common, and (L-R) is the current on triphase L and -R (mostly, as I noted above). We're saying the same thing physically, only the interpretation is different. What you've computed *is* the triphase effect, not just a sort of min/max/mono thing that approximates it, and I would not say that a triphase L and R are not very different. They are vitally different.

Yes and no.

Triphase files can include a lot more than just phase shifts. Most of the files I produce have reduced amplitude when the signals are in phase, to prevent the common electrode feeling much stronger than the others. If I've got time, I sometimes go a step further and reduce one channel or the other independently so that different strokes have different sensations. Sometimes I'll entirely mute one channel so that there is effectively a mono signal between the common and a single electrode. This is what I mean about differences between L and R - when the amplitude or frequency are significantly different, rather than just slight phase-shifts at a constant amplitude.

Converting a triphase file (even a simple one) to single/dual channel doesn't preserve the triphase effect. The reason that triphase produces the stroking sensations is because the flow of current gradually shifts between L to R and L+R to Common. The sensation moves gradually backwards and forwards, instead of fading on and off. Summing the signals together computes what happens at the common electrode but it loses the interplay between the electrodes. Even fading one channel up as the other fades down does not give the same intermediate sensations where some of the current flows to the common and some of it flows to the opposite electrode. The resulting feeling is very different.

This isn't something I stumbled upon by accident. My method of converting triphase files to single/dual channel stems from deliberate steps that I take when creating/adjusting estim files in audacity. You don't need to explain to me how it works!

[zebbg69]

Triphase files can include a lot more than just phase shifts. Most of the files I produce have reduced amplitude when the signals are in phase, to prevent the common electrode feeling much stronger than the others. If I've got time, I sometimes go a step further and reduce one channel or the other independently so that different strokes have different sensations. Sometimes I'll entirely mute one channel so that there is effectively a mono signal between the common and a single electrode. This is what I mean about differences between L and R - when the amplitude or frequency are significantly different, rather than just slight phase-shifts at a constant amplitude.

This is all true and is all literally the triphase effect. Current running out of each electrode is physically the sum of the other two. For symmetric resistances & volume knobs, that puts L+R at the common and 2L-R at the "A" and 2R-L at the "B." It's just electricity, Ohm's Law and Kirchhoff's Law in a closed loop with your body in the middle mediating the current--it does not matter whether the creator thought about phase or tweaked volume levels or even used any sinusoidal forms to make the file. For actual electrode placements (resistances) and volume knob positions, each L and R term gets scaled by some factor, so it's really aL+bR 2cL-dR, etc. for empirical a,b,c,d, but the overall feeling of it is captured by L+R and L-R, and you can always tweak those proportions when computing it if desired plus the volume knobs give you some control over that after the fact.

Of course you are right that "L+R" is not the whole effect--because that's only one electrode. You need the L-R element to get the opposite side of it and complete the picture. The "B" electrode is basically just the complement of the "A" so there's no additional information there.

This is not an argument. I am helping to answer someone's prior question as to whether to take L-R or R-L and why, in conjunction with L+R. The answer is you only need one, and you can pick L-R or R-L arbitrarily, pair it with L+R, and the reason is physics. Draw a triangle with one-Ohm resistors, label the vertices A, B, and Ground (Common), and then calculate the currents. That's all I did above. There is no other hidden "triphase" action to be analyzed. It's just the currents flowing when we tie those common leads together. If there are concerns about frequency response, that is already factored in. I'm only composing the currents that already result from flowing in the body, before tying the leads together vs. after. Just tying the leads together does not activate some new "phasey" branch of physics. We're just summing together whatever was already flowing when not tied together. Phase response would not change the math either, and if it did then the phase modulated "stroker" technique would not work. It only works because the phase carefully put into these waveforms does reach each electrode undistorted to produce the intended interference that makes the stroke move up & down.

I'm not knocking anyone's technique or prior claims, just showing why (L+R, L-R) captures the triphase effect for dual channel at the point where it was initially asked in this thread. I have a master's degree in this stuff, so I am not just pulling it out of my ass like some "B" electrode. I love your work, LondonGent, and have no desire to argue with you, and I allow that you may have arrived at your results with other heuristics, and it has all worked great and everyone is happy with the results. I am just relaying what they teach in electrical engineering departments that applies to this.

[edger477]

There is no other hidden "triphase" action to be analyzed.

one thing besides the technical aspects... sensation is interpreted in brain, and for some even the "triphase" might not be perceived as it is without accompanying video. With conversion to DC, it is even harder to make a setup where sensation gradually switches in intensity from one part of body to another, but with something in visual spectrum to trick you it might be possible.
Whether conversion works or not will mostly be dependent on any particular brain's ability to interpret the sensations in desired way.