Topic: expand stereo image

[panther65]

Hello-
I have a few nice soundboard recordings that are stereo, but quite narrow in terms of the soundstage. Is there a plugin or a setting that would work best to move the images into a more spacious stereo spread.
I have tried the pan/stereo expander is adobe audition 3 and that works ok, but gives an unnatural phasey quality to the vocals.

[boojum]

This is an old SE's trick you can try with the VoxEngo MS decoder/encoder.  Encode the stereo into MS and then decode it back to stereo.  On the decode back you can vary the ratio of side to mid which may help. The audio editor I use, SAM 10 Master, has this function built in.  Perhaps the SAM 9 SE does also, I do not know.  It has a trial period of 30 days but I do not know if it is fully functional.  I think it is.

Good luck.

[panther65]

Thanks a lot for the plug-in idea. I downloaded it and am using it with adobe audition and sam 10. Mainly use audition because I haven't found a good way for sam to work with files over 2gb long.
Anyway, it has an inline option that seems to accomplish what I'm after. However, would it be better to encode to MS?

[OFOTD]

However, would it be better to encode to MS?

No.   You don't have a mid-side recording in the first place.  Your final output would have no resemblance to the original sound.

Not sure about Audition or SAM but SoundForge has a pan/expand plugin.    Also consider the free trial of Izotope's Ozone 3 quite a bit of really good plug-ins that would help simulate what it appears you're trying to accomplish.

Stay away from trying to convert to m/s and back.  Not good.

[Brian Skalinder]

Mainly use audition because I haven't found a good way for sam to work with files over 2gb long.

Are the original source files > 2 GB?  And are you working in SAM's VIP or WAV editor mode?

IME, SAM SE v9 doesn't like source files > 2 GB.  But that's no problem - I have my recorder set to limit file size to ~2 GB.  SAM's VIP (project) mode then allows me to join multiple files together seamlessly and apply my edits across all source files at once (as though I was editing a single file).

[boojum]

The newer releases of SAM handle way over 4GB files effortlessly.  As for going from stereo tracks to MS and back being bad, well, I had not heard that before.  Interesting.  I have no personal experience to support that.

[OFOTD]

As for going from stereo tracks to MS and back being bad, well, I had not heard that before.  Interesting.  I have no personal experience to support that.

Weren't you suggesting that he do that here though?

This is an old SE's trick you can try with the VoxEngo MS decoder/encoder.  Encode the stereo into MS and then decode it back to stereo.  On the decode back you can vary the ratio of side to mid which may help. The audio editor I use, SAM 10 Master, has this function built in.  Perhaps the SAM 9 SE does also, I do not know.  It has a trial period of 30 days but I do not know if it is fully functional.  I think it is.

Good luck.

I'm confused.

[Brian Skalinder]

Stay away from trying to convert to m/s and back.  Not good.

Why?  It's just simple math (i.e. addition and subtraction)^*.  Theoretically, one could could convert back and forth (encode stereo to M/S, then decode back to stereo) 3 times, and the end result should match the starting point (assuming no other operations are performed that would introduce other changes, like SRC, dither, noise shaping, etc.).  I just processed a sample for kicks:

• Take SAMPLE.WAV (a stereo WAV file).
  
• Encode SAMPLE.WAV to M/S and save as SAMPLE_MS-ENCODED.WAV
  
• Decode SAMPLE_MS-ENCODED.WAV to stereo and save as SAMPLE_MS-ENCODED_MS-DECODED.WAV
  
• Sum SAMPLE.WAV and inverted SAMPLE_MS-ENCODED_MS-DECODED.WAV
  
• Results:  NULL waveform

Granted, I only did it once - not three times.  But I don't see how performing the same process multiple times would yield a different result.  Perhaps I misundersand the process...  ??

^* Encoding stereo to M/S and decoding M/S to stereo really is "just (simple) math".

Decoding M/S to stereo

L = M + S
R = M + (-S)

Encoding stereo to M/S

M = L + R
S = L - R

[mr qpl]

I've had decent results using the graphic eq on each side a little differently. As with most of these methods, I'd "save early and save often"

[Will_S]

I think OFOTD is saying that encoding R-L to mid-side (or sum and difference) and then tweaking sum vs. difference before converting back to R-L is unlikely to produce good results, not that it is theoretically impossible.

Without having tried it much myself, I'd imagine that L-R > sum-difference > L-R should work well for altering the stereo width of recordings that are based entirely on intensity differences (i.e.., coincident micing techniques) but might not work so well on recordings that get some of their imaging from timing and phase differences as well.  But I read a paper from a guy saying that converting a stereo recording to sum and difference, then boosting the bass in the difference channel before going back to L-R, was the bees knees for recordings made with cardioids at 5cm and 120°.  But come to think of it I think he recommended against that approach for more widely spaced mic pairs, while other folks swear by it for XY and MS recordings.

edit:

As for going from stereo tracks to MS and back being bad, well, I had not heard that before.  Interesting.  I have no personal experience to support that.

Weren't you suggesting that he do that here though?

I read this as saying he didn't have any experience suggesting it was a bad thing, not that he'd never done it.

[Brian Skalinder]

I think OFOTD is saying that encoding R-L to mid-side (or sum and difference) and then tweaking sum vs. difference before converting back to R-L is unlikely to produce good results, not that it is theoretically impossible.

Encoding stereo to M/S and decoding back to stereo while tweaking the M:S ratio strikes me as a very good way to adjust the stereo image.  Agreed, though, with your comment that it's likely to produce best results with intensity difference stereo imaging.

[live2496]

I think OFOTD is saying that encoding R-L to mid-side (or sum and difference) and then tweaking sum vs. difference before converting back to R-L is unlikely to produce good results, not that it is theoretically impossible.

Encoding stereo to M/S and decoding back to stereo while tweaking the M:S ratio strikes me as a very good way to adjust the stereo image.  Agreed, though, with your comment that it's likely to produce best results with intensity difference stereo imaging.

Mastering engineers do this all the time. I've never heard any warnings about doing this other than being careful not to apply some equalization that causes a phase shift. If eq'ing I would apply wideband eq (low Q) to it.

But I don't see why you couldn't even just compress the mid signal and leave the sides alone and leave the stereo part unaffected by the compression. Then blend the mid/side together again. Mid-Side processing opens up a lot of possibilites.

In Samplitude, the stereo control all the way to the right (enhance) is the side signal. All the way to the left is the mid signal. Lately I have been bouncing mid and side to separate stereo tracks and processing separately on some audio that I have been working with. It doesn't sound weird to me in any way, but I suppose you could get too crazy with it. Like if you have too much side signal and not enough mid. When you use too much side the recording starts losing the stuff coming up the middle which is usually lows and bass.

On the last recording I worked on it was a dry track with voice and guitar recorded in stereo.  I applied a little reverb to the sides only and I thought it turned out pretty nice. If you add reverb to lows things get sounding unclear and muddy. This is why outboard reverbs sometimes have a high pass filter. So adding reverb to the side signal works. Pretty interesting stuff really.

Gordon

[boojum]

I think OFOTD is saying that encoding R-L to mid-side (or sum and difference) and then tweaking sum vs. difference before converting back to R-L is unlikely to produce good results, not that it is theoretically impossible.

Encoding stereo to M/S and decoding back to stereo while tweaking the M:S ratio strikes me as a very good way to adjust the stereo image.  Agreed, though, with your comment that it's likely to produce best results with intensity difference stereo imaging.

Mastering engineers do this all the time. I've never heard any warnings about doing this other than being careful not to apply some equalization that causes a phase shift. If eq'ing I would apply wideband eq (low Q) to it.

This was my initial point but somehow I think it was misunderstood.  I suggested it as a way to tweak the stereo a bit in an editor which did not have the ability or a plugin to do that.  I had also read it was common or know practice.  Makes you kind of want to ignore requests for help around here.

And to make the other statement clearer, no, I have not had bad luck doing this nor have I heard of anyone else having bad luck.  It is a common practice with professionals.  That is why I thought the comment that it was a bad practice interesting.

[DSatz]

Sorry to arrive late. The trick of converting an L/R stereo recording to M/S, then boosting (or otherwise playing with) S and rematrixing back to L/R can be very successful with certain recordings but much less so with others.

You can usually get an idea of how well it's likely to work by listening to the recording in mono. If the recording becomes muffled or muddy or you lose much bass when the channels are summed, then the trick is not likely to work as well as if the recording simply collapses to a center source image while keeping the same overall sound quality (timbre) as the stereo recording had. The best outcome will normally occur if the original recording was a two-mike coincident recording (since those are inherently mono-compatible), and somewhat less so if it was near-coincident (e.g. ORTF), but expect crap if spaced omnis were used.

In the coincident case, converting to M/S will give you a pretty close approximation of what the recording would have been like if an M/S microphone setup had been used in the first place. And if that's what you have, there are other options besides simply boosting the S channel to increase the difference between the eventual left and right signals. You can also apply EQ--I'd suggest at least trying a low-frequency boost to see whether you like what that does. Yes, too much of it (and reaching too high up in frequency) will cause "phasiness," but a moderate amount (however you may choose to define that) won't necessarily do so.

There's always a limit to how much trickery and deceit you can use before you risk producing an "unnatural" sound quality that's more distracting than it is disappointing. Unfortunately, the way things seem to work out most of the time, a pretty good recording can often be made to sound quite good, while a mediocre or poor recording may well resist being improved by very much. An old friend of mine used to call that the "law of conservation of goodness."

If the M/S approach doesn't work, a "stereo synthesizer" can be used (formerly outboard processing units, but often a feature of audio editing software nowadays). Similarly, moderate use of a good stereo reverb (or reverb software) can help add a sense of space even if there's nothing particularly true-to-life about the space that it adds a sense of. Both techniques were widely used in LP mastering, and even in some CD mastering though I probably shouldn't admit that.

--best regards