Review – Two Mono-to-Stereo Converters
Imparting “stereo-ness” to a mono track can be done a variety of ways, some of which make subsequent playback in mono problematic. Here are two plug-ins that deliver stereo with none of the typical problems.
by David Baer, May 2015
We are going to look at two plug-ins that nicely convert mono tracks into an interesting stereo sound: MStereoSpread from MeldaProduction and Spatifier from Voxengo. They are both worthy contenders for your attention and get the job done brilliantly, although they approach the problem in very different ways.
It’s sometimes beneficial to take a mono track and add some width to it by introducing a pseudo-stereo quality to the track. Doing so requires finding some way to make the left and right tracks slightly different. This can easily be done by using delay, reverb or other tricks. For many producers, that’s enough – no further concern is warranted that the resultant track might sound degraded if played back in mono. But to those who want to preserve mono-compatibility, care must be taken. Some stereo-simulation tricks can introduce phase problems that will do nasty things when the stereo is reconstituted as mono. I wrote about this at some length here:
To summarize what was said in that article, a mono signal on a stereo track has identical signals in the left and right channels. When producing pseudo-stereo, mono-compatibility can be maintained if we add a signal to one channel and subtract the exact same thing from the other channel. If left and right are later summed to create mono, the alterations cancel each other out and the listener is none the wiser. Any stereo-ization technique that does something else cannot be said to be mono-compatible, although we can minimize the risk of such problems with appropriate techniques.
So, in what ways can we alter identical left and right signals that will produce acceptable stereo? How about just adding/subtracting the original signal itself? Obviously, this is no good as it will just make one channel stronger and the other weaker. We’re just affecting the L/R balance.
OK, how about using, say, an attenuated mono mixdown of Pinball Wizard from Tommy? Again, obviously this is not going to produce satisfying results. Clearly, whatever we use must be closely related to the original signal. So, what are our options?
We have several, the first of which we’ll consider is to take the original signal and moderately amplify some bands of the audio spectrum and moderately attenuate others. When we add the resultant signal to one channel, the amplified bands get louder and attenuated bands softer. When we subtract from the other channel, it’s the attenuated bands that get louder and amplified bands softer. If we have a sufficient number of bands and choose them wisely, we can get a pleasantly wide result that still sounds centered. The sense of wideness will be governed by the degree of amplification/attenuation, the more of the latter, the greater the wideness.
Care must be taken, however, not to alter too-low frequencies in this way or we actually might get a lopsided result, a shift in the perceived audio center. If we stick to just the overtone area of the spectrum, the presence and air territories, we’ll be completely safe. Even if we dip down into the range of the fundamental frequencies, we can often achieve satisfactory results. But the really low frequencies should normally be left alone.
A second technique can also be effective. Take an attenuated copy of the original signal and delay it somewhere between 5 and 50 milliseconds. Add it to one channel and subtract from the other. The less the attenuation (i.e., the louder the delayed signal), the wider the result will be.
This approach has a couple gotchas however. First, we do not want too-long a delay or it will be heard as an echo. Keeping things under about 60 milliseconds will avoid that. What we are doing in applying the delay is introducing comb filtering into the respective channels. Since it’s complementary, it does not become an issue with mono-compatibility – the alterations cancel each other out upon averaging. However, the comb notches may occupy not only fundamental frequencies but harmonics as well. Thus, the instrument making the sound may appear to be unstable L/R-position-wise. Tuning the delay amount (effectively tuning the fundamental notch frequency) on a case-by-case basis can be quite helpful in minimizing this problem.
A third technique is similar to the delay technique. Rather than a delay, use something like a mono reverb signal to add-to/subtract-from the respective channels. The signal alterations will be closely related to the original signal and the result can sound natural and unobtrusive. But the risk we run is in imposing an unwanted depth to the results (which, by the way, can also be a corollary effect of the delay approach).
Near (but not absolute) mono-compatibility may be achieved by adding/subtracting something other than the same signal. If two uncorrelated reverb-like signals are created and applied, the possibility of unwanted comb filtering exists, but the chance is minimized courtesy of the lack of correlation. In this case, if the resultant stereo is down-mixed to mono, the residual effect is closer to that of a mono-reverb with a fairly short tail.
We can get reverb via algorithmic methods or through convolution. With convolution, we essentially apply a series of filters of many bands to each sample, the processing occurring for the length of the reverb tail. The data controlling the process is called an “impulse”, and convolution reverbs use impulse data gathered from measurements of real-life venues.
But an impulse can also be synthesized and used for convolution. Significantly, it does not need to be a reverb process per se. For starters, it does not have to mimic real-life locations. Certainly, it can be created to gently alter left/right signals expressly in order to add width. With a carefully-devised algorithm for creating the impulse, adding a sense of depth can be achieved or avoided as desired (I am by no means suggesting that this is trivial, rather just that it’s possible). But in the end, it does not need to be a reverb impulse at all. Something closely related to the original signal will do. When altering the original, if just the right amount of difference between left and right can be introduced in a way that does not compromise the original content, the goal of pseudo-stereo has been achieved.
MStereoSpread from MeldaProduction
So, let’s look at the first of our two candidates for mono-compatible stereo-ization, MStereoSpread. This plug-in implements the first two of the techniques explained above. They may be used individually or together. A small number of factory presets ably present the range of capabilities.
The spectrum generator module implements the spectrum-band alteration technique and is quite straightforward. Spread controls the amount of the effect, Bands the number of spectral bands, and Min Frequency the frequency below which no separation will be introduced. Focus controls the emphasis of lower or higher bands in terms of the extent of amplification/attenuation. Positive values dictate that more expansion is to happen in the higher bands. Finally, Saturation does just what it says, causing higher harmonics to be generated in the added-in signal.
Invert simply flips the band emphasis in the other direction. This can be quite useful when running MStereoSpread on multiple tracks using the same settings. Mixing up the band emphasis can produce a more-rounded, evenly-distributed result.
The delay-based generator is almost as straightforward; it implements the second of the techniques described earlier. Spread, Min Frequency, Saturation and Invert do the same things as in the spectral generator. The delay length is actually specified with a frequency. You may find it annoying to not have a simple millisecond value to work with here, but frequency is actually an equivalent parameter and may be far more convenient to work with. Delay will introduce left/right differences courtesy of comb filtering. If we just avoid the predominant frequencies in the musical content, we will likely be happier with the result. On the other hand, we could just call this one a “set by ear” parameter – just tweak until it sounds good.
The delay module actually supplies two delays. If you wish to use the second one, the Secondary Depth parameter can be set to higher than 0% and the Secondary Ratio supplies a multiplier of the primary frequency to arrive at the secondary frequency value.
This being a MeldaProduction plug-in, there is of course much more. The phase scope provides useful visual feedback of the processing. But the really interesting possibilities come courtesy of the modulation options. I won’t bother to detail how these work – they are rather sophisticated after all. But I just want to mention the possibilities as a special effects unit they introduce. A natural stereo enhancement is what will normally be desired. But by making the separation parameters fluid, some fairly exotic and/or spacey results may be achieved.
Let us now turn our attention to Voxengo’s Spatifier, which implements the third of the techniques mentioned above. The UI may look somewhat other than obvious but its operation is fairly easy to understand. The plug-in generates a pair of synthesized convolution impulses based on a variety of user settings. These are then applied to the original signal producing a distinct left and right channel signal.
The user settings have several parts. The top level of sixteen controls in the upper array of sliders allows per-frequency-band specification of the timing of the convolution impulse for each band. The values range from 1 to 250 milliseconds.
The Impression control dictates the overall sense of depth in the generated convolution impulse. The choices include six settings: Closest, Very Close, Far, Very Far and Farthest. It adds a depth quality to that already being governed by the band-specific timings.
The bottom level slider controls dictate how much of the convolution is to be applied in each band. It is the dry mix amount. At 100%, the signal is unaltered (fully dry). At 0%, the convolution is fully applied. The Dry Mix knob on the right then controls the overall Wet/Dry ratio.
Finally there’s the Random Variation control. The convolution impulses are synthesized and this control provides a randomization seed for the calculations. Different values can produce decidedly different results. I would say that this is a “set by ear” control, except that you might be there all day fiddling with it since there are so many different possibilities.
The amount of variation in the stereo results produced is quite impressive. You can achieve a very up-front result with random variations of the Closest Impression, rather deep results using the Farthest Impression, and anything in between. You can easily restrain the processing to upper portions of the audio spectrum or not. Once you get a handle on how this plug-in works, it’s completely straightforward to come up with your own programs that do just what you need.
Are Either (or Both) of These Plug-ins for You?
Let’s start with the good news. Both of these plug-ins are superb. MStereoSpread creates pseudo-stereo from mono brilliantly. Spatifier creates pseudo-stereo from mono brilliantly. Of either we can say: it does its job with aplomb and panache. The results produced are invariably credible and natural.
So what’s the bad news? The bad news is that they are distinctive enough in character that you are likely going to end up wanting both. It’s not a case of which one is superior. They are both first rate, but they each have a distinguishing quality about them.
There are a few differences in capabilities. MStereoSpread has the convenient Invert control. However, you could put Voxengo’s free MSED plug-in after Spatifier to flip the channels with minimal extra effort. MStereoSpread has modulation capabilities not available in Spatifier, although I doubt most users would take advantage of these in other than rare occasions.
So, it comes down to demoing both yourself, and I highly recommend taking the small effort to do so. They are close enough in price that price should not play a major factor in the decision. MStereoSpread lists for for €79 EUR, but MeldaProduction has rotating sales that will periodically make it available at half that amount for a brief period. Spatifier lists for $70 USD, but discounts are available to return customers or those purchasing multiple products in one order.
For more information or to purchase these plug-ins, go here: