Clarification on True Stereo for convolution reveb (Official feedback from Propellerhead)

[carlosedp]

Hi guys, base on recent discussions, I submitted a question to Propellerhead regarding true stereo support on RV7000 MK2. As seen below, Fredrik pointed me out that the true stereo algorithm is not applied to convolution so the two reverb technique (in a combinator) must be used:

Hi Carlos,

Thanks for pointing this out.
The "true stereo reverb" phrase is true for all reverb algorithms except for Convolution. I have now clarified this in the documentation which will be released with the program.

You are also right about the fact that you will need two RV7000 MkII devices if you want to create a true stereo convolution effect. I think there will be some Combinator Effect patches for this by the time of release.

Best regards,
Fredrik Hylvander
Propellerhead Software

Carlos

[ahs]

that means only mono in and stereo out ?

[carlosedp]

Actually a non true stereo reverb merges the input channels (sums to mono) than applies the reverb algorithm generating a stereo output based on the IR.

To have a true stereo reverb, you must process using two sets of IRs, one stereo for left and one stereo for right.

This picture illustrate the flow:

 


This is only valid if you have two sets of IRs per preset, like the M7 IRs from Samplicity. In case you only have one stereo IR (like 480L from BigG), you don't need this setup.

Carlos

[selig]

Actually a non true stereo reverb merges the input channels (sums to mono) than applies the reverb algorithm generating a stereo output based on the IR.

To have a true stereo reverb, you must process using two sets of IRs, one stereo for left and one stereo for right.

This picture illustrate the flow:

 


This is only valid if you have two sets of IRs per preset, like the M7 IRs from Samplicity. In case you only have one stereo IR (like 480L from BigG), you don't need this setup.

Carlos

There's no need to use two cables to each reverb, since the idea is to use one channel of each (since both devices are mono).

[ScuzzyEye]

There's no need to use two cables to each reverb, since the idea is to use one channel of each (since both devices are mono).

Yeah, one cable in, two cables out.

To think of it another way. Each side of the source would be it's own point in space. They will each radiate sound outward, and be affected slightly differently. So they should be convolved on their own. The convolution is also stereo, so each single point becomes stereo, and those two sets of signals are mixed back together.

[selig]

There's no need to use two cables to each reverb, since the idea is to use one channel of each (since both devices are mono).

Yeah, one cable in, two cables out.

To think of it another way. Each side of the source would be it's own point in space. They will each radiate sound outward, and be affected slightly differently. So they should be convolved on their own. The convolution is also stereo, so each single point becomes stereo, and those two sets of signals are mixed back together.

One cable in, one cable out - If the device is mono (loading one channel in each device), no need for the second cable, right?

[ScuzzyEye]

One cable in, one cable out - If the device is mono (loading one channel in each device), no need for the second cable, right?

That depends on if the IR is stereo or not. I think there were some pairs of stereo IRs (or quad channel for supported convolution effects).

[selig]

One cable in, one cable out - If the device is mono (loading one channel in each device), no need for the second cable, right?

That depends on if the IR is stereo or not. I think there were some pairs of stereo IRs (or quad channel for supported convolution effects).

Good point - I was assuming stereo IRs. For Quad channel support why not use four devices? (other than CPU issues)?

[ScuzzyEye]

Good point - I was assuming stereo IRs. For Quad channel support why not use four devices? (other than CPU issues)?

The CPU usage is likely to be very close to the same. The dual stereo IRs are distributed as quad channel waves, they seem to be designed as dual stereo, not quadraphonic. Then they are also packaged as 2 stereo waves, for greater compatibility. I haven't seen them as 4 mono. So that's the primary reason why not.

[ScuzzyEye]

Here's an example to get an idea of how the dual stereo IR are made.

Imagine a piano on a sound stage. An IR is made from the low end of the sound board, and picked up by a stereo mic pair. Then a second IR is made at the high end of the sound board, and picked up by the same mic pair.

You can then convolve a stereo signal with both IRs to get a similar response to how stereo source would sound in that environment.

[selig]

Good point - I was assuming stereo IRs. For Quad channel support why not use four devices? (other than CPU issues)?

The CPU usage is likely to be very close to the same. The dual stereo IRs are distributed as quad channel waves, they seem to be designed as dual stereo, not quadraphonic. Then they are also packaged as 2 stereo waves, for greater compatibility. I haven't seen them as 4 mono. So that's the primary reason why not.

Got it - I have no real experience with the quad files, so I was basically speaking from ignorance there - sorry about that, folks!

[Ostermilk]

Basically true stereo involves two stereo IR's taken usually in different parts of a room.

Check this for the schematics of the different types of playback (provided you have a true stereo pair of IR's...i.e two two channel pairs taken within the same space):
http://www.liquidsonics.com/software_re ... stereo.htm

So going by the middle diagram you can see that in reason you will need two RV7000's wired in TS mode to use the pair of IR's.

Another way rooms and spaces are commonly captured is by using the B-Format Ambisonics 4 mic (WXYZ) method, you can use these results through a true stereo reverb which won't give you an ambisonic result but can it be effective in making the resultant reverb more lush nonetheless if you come across IR's in this format.

Here's a simple TS combi with an optional touch of chorus for some movement and knobs to control both RV7k's at the same time. Just go find some true stereo impulses to use as none of it makes sense unless you've got an A and B pair of two channel files to use in it. Edit: I've now include a TS pair of IR's I just cooked up too.

TrueStereo Convo (2).zip

[Ostermilk]

Actually a non true stereo reverb merges the input channels (sums to mono) than applies the reverb algorithm generating a stereo output based on the IR.

To have a true stereo reverb, you must process using two sets of IRs, one stereo for left and one stereo for right.

This picture illustrate the flow:

 


This is only valid if you have two sets of IRs per preset, like the M7 IRs from Samplicity. In case you only have one stereo IR (like 480L from BigG), you don't need this setup.

Carlos

There's no need to use two cables to each reverb, since the idea is to use one channel of each (since both devices are mono).

Both devices have different stereo IR's though so the output from each of the 4 outputs is different regardless of the fact that each RV7000 is being sent a single split channel to each of its inputs, hence the cabling is correct. Carlos is absolutely correct here. Note though that 'Parallel' mode needs to be selected for this to work as per the TS spec or you may as well just use single inputs to each RV7000 as the Reason device does sum the inputs to mono in every other mode.

See the earlier schematic as well as the middle one (compared to the first standard single stereo IR Parallel mode one) on the Liquidsonics page, the combi and example files in previous post for further clarification of why the signal path works this way. Basically both channels will be affected by the corresponding L+R channels of both two channel IR's this way.