Surround Sound From Stereo

Waves S360 Surround Imager

More and more material needs to be delivered in surround formats, but what if you don’t have the time or the budget for a 5.1 remix? We test the current crop of ‘unwrap’ plug-ins that can create a surround master from a stereo mix.

The best way of creating a surround master is to mix, or remix, from the source multitrack session into surround. However, this isn’t always possible, either because the source material is not available, or time and budget don’t permit. This is where ‘unwrap’ plug-ins can come to the rescue. In this feature, we are going to take a look at the ‘unwrap’ plug-ins currently available for Pro Tools, try them out on a range of different test material, and see what they can and can’t do. As you’d expect, all of these plug-ins require an HD system to run, though Cycling 74’s Upmix runs natively rather than on the HD cards, and TC’s Unwrap is also available in Powercore format for other platforms.

The Art Of Unwrapping

When undertaking unwrapping, there are a few things you need to consider: whether you want the process to be purely additive (in other words, creating new content for the Rear and Centre channels but leaving the main L and R speakers untouched), whether the ability to collapse to stereo again is important, and whether an LFE channel should be derived or not. All of these requirements will affect what processes you can apply to your source material without experiencing any unpleasant side-effects, and there are also some general rules that should be followed where possible:

• Try to put the ‘unwrap’ plug-in as the last plug-in in the chain on a track. This will be the most DSP-efficient way of working, as it means that all your other plug-ins will be processing either mono or stereo audio rather than five- or six-channel audio.
• Do not rely on being able to create precise sound images placed between speakers at the side of the listener for stationary sounds. Our ears are very poor at localising sound to the side, so while movement in sounds through the sides is fine, reliable placement of sounds between, say, the Left Front and Left Rear speakers is impossible.
• Heavy use and manipulation of the Centre channel can hinder the accurate folding down of the mix to stereo. If proper fold-down is important, make subtle use of any Centre channel processing, and always check your mix for compatibility.
• The same goes for the Surround (Rear) outputs: the more processing you employ, the more any subsequent ‘downmix’ will be compromised.

Whenever you are mixing in surround, in whatever format, downmix considerations should be always in your mind. You cannot predict how your surround mix is going to be listened to further on down the chain. Where possible, it is best practice to include a properly remixed stereo version of any surround content. The DVD-V and DVD-A formats both allow for this. However, it isn’t always possible or doesn’t always happen, so you have to consider that some listeners will hear your surround mix in stereo or even in mono derived from a downmix, created in their equipment, of your surround mix. So when mixing or upmixing in surround, you need to regularly check how your surround mix sounds when it is downmixed again.

The Contenders

• Waves S360 Surround Imager

Waves’ S360 is a panning and imaging tool for 5.0 or 5.1-channel surround, which can be used to unwrap both mono and stereo sources to either 5.0 or 5.1 surround formats. It lets you set Rotation and Width for a mono, stereo, or surround source in the surround mix, and add Room Model Early Reflections for distance positioning and Shuffling for enhanced low–frequency width.

You can set Rotation (panning) angle for the centre of the source image manually, or use the mouse to grab the red dot on the graphical display. The rotation control can be set to work Pair-wise or Triple-wise. The former works well between two points, even if the listener is not exactly in the centre of the sound stage, and tolerates poorly set–up sound systems, as the phantom image will still appear between the two speakers. The Triple-wise pan-pot, by contrast, uses three speakers to localise the phantom centre. This can provide a more stable phantom image, but relies on the playback system having very similar speaker positioning and types to your monitoring system; the Pair-wise pan pot is more reliable across different speaker setups. You can also specify directly how much the Centre speaker should be used.

The Width pan-pots are used to spread the sound source. If you’re working from a stereo source, four different Width pan–pot configurations are available: Mono Divergence, Balanced, Front-stage and Focus. Mono Divergence spreads the energy of the mono source outwards from its indicated direction, such that it will be almost equally loud in all speakers. The Rotation will still preserve direction, albeit much less coherently. It behaves in a similar way on stereo sources, except that it increases correlation of the sound between speakers. The sound will still be mono, but it will be less localised at its indicated Rotation and in the extreme width.

The Balanced width pan-pot spreads the image to produce a wider front while narrowing the rear. In effect, it will widen the stereo stage, apparently shifting the side phantoms further to the sides and backwards. Front-stage width panning widens the front stage by multiplying the directional image width by the specified ratio, while the Focus pan-pot takes each speaker pair and extracts the Middle and the Sides signals. A value of zero for Width focuses only the ‘M’ signal, and a Width value of 4 only the ‘S’ signal. This allows you to balance direct against ambient signal, with a value of 4 representing all ambience, while zero is all direct.

The panning graph in the centre of the plug-in window displays a splined-circle energy scope, which is another way of representing the information that appears in the bar-graph meters.

The second main element of Waves’ ‘unwrapping’ technology is the addition of early reflections to give the unwrapped signal a sense of being positioned in a space. Balancing the direct sound and Room Model Early Reflections creates ‘distance’, and the specified room size affects the sound of the reflections. The range of distances available is linked to the room size by default, but this link can be broken to allow imaging closer inside the room or to adjust for unnatural settings, to make a source sound more distant than is ‘really’ possible in that room. To complete the room emulation, you may want to send the channel’s output to a surround-capable reverb plug-in.

The third key element is Shuffling, an enhancement process which is claimed to add a sense of spaciousness and life to an otherwise ordinary spatial image. The normal stereo effect from two loudspeakers creates the illusion to the listener of phantom images lying between the loudspeakers, but it is not perfect, with the apparent stereo stage width narrower at bass frequencies than in the mid and high frequencies. The reason for this is that the ears and brain determine sound direction by different methods below and above around 700Hz. Shuffling is a process that widens the bass frequencies, with the phantom bass and treble images seeming to be of comparable soundstage width when the shuffling level is set at around 1.6. Higher settings will create less precise stereo images, but generally tend to produce an illusion of ‘spaciousness’. Shuffling has no effect on the central imaging, only on sounds panned away from the centre, so it does nothing to mono images, but can enhance the stereo effect already present in stereo images.

• Cycling 74 Upmix Cycling 74 Upmix.

Upmix from Cycling 74 uses their Pluggo technology, which means that the Upmix installer will put the plug-ins in the Library / Audio / Plug-ins / VST / Pluggo folder and you won’t find it-in in your Digidesign Plug–ins folder. Instead, for RTAS support a ‘stub’ plug-in called Pluggo-RTAS is installed there. This is the mechanism that Pro Tools uses to gain access to the Upmix plug-ins.

However, this causes a problem with routing to the Upmix plug-in in a Pro Tools environment. The plug-in has to be inserted on a 5.1 track or aux input, and you can’t insert it on a mono or stereo track as you would expect. So, to get it to work, you need to create a stereo aux input and route its output to a surround bus which the Upmix track can pick up. Also, there are 23 presets to start from, but because you select the preset from the View drop-down menu, the display doesn’t show which preset you are working on. The only workaround I could come up with was to save each preset as a Pro Tools preset from the proper preset menu.

Upmix is actually a suite of six plug-ins. The flagship plug-in achieves the actual ‘unwrapping’, with a number of algorithms and positioning controls that we will be looking at here. The five additional plug-ins handle other common surround processes: Rotator is a six-channel panner, Reroute rearranges the channel order to conform to various surround formats, FoldDown allows you to check your surround mix in stereo, with appropriate centre- and surround–channel gain reduction, Rebalance is a simple set of level controls to adjust each of the surround channels and LFE is a low–frequency effects generator.

The main Upmix plug-in is divided into a number of sections. The Centre Module creates a gain-adjustable mono signal by summing the left and right channels of a stereo source, with optional low-end roll–off and a single-band parametric EQ. These allow you to filter out low frequencies from the centre-channel output and/or emphasise or reduce other frequencies; you could, for instance, put a presence peak in to improve the intelligibility of vocals.

Upmix cannot be used on a conventional stereo channel; instead, it must be placed on a surround track or auxiliary.You can choose one of five Surround Process algorithms to control the generation and tuning of the surround channels. The Ambience algorithm extracts the stereo ambience from the front source and sends it to the rear channels using true stereo rear–imaging, and is the best process to use if reliable stereo fold-down is a consideration. In the Immersion algorithm, by contrast, the left source channel is split into frequency bands to the left and left-surround output channels, with the same process being used for the right source channel. This algorithm is designed to work well with mono sources. Spread is a more aggressive version of the Immersion algorithm, which can provide a wider feel but is less fold-down compatible, while Quad is a four-channel version of the immersion process, and maintains a non–processed centre channel for situations where the centre channel is exclusively for dialogue. Multidepth is an even more aggressive variant of the Immersion process, and Cycling 74 warn that it is not fold-down compatible. Finally, in Mirror mode the front channels are simply duplicated to their surround equivalents, which can be useful when used in conjunction with the individual channel delays provided in the Output Levels section, though, again, delays of more than a few milliseconds will interfere with stereo fold-down.

The LFE generator is designed to derive, from the input signal, a low-end signal suitable for use as an LFE channel. The input source is high-pass filtered and can also be gated before reaching the two processes that actually generate low-frequency content. The first of these is an Oscillator, which allows the addition of a static sine- or square–wave tone in the LFE channel. The Oscillator relies on careful adjustment of the Gate section for triggered modulation of the static tone, and accurate setting of the high-pass filter to isolate the signals you want to trigger the gate. The idea is that you can have it fire only on certain elements of the original source, such as kick drums, explosions, engines and so on. The second low-frequency generator, the Downshift section, creates LFE material by pitch-shifting the gated and filtered input signal, though it uses a lot of DSP power, so it is best to turn the Downshift section off when you don’t need it.

Finally, the LFE output can be low–pass filtered again, and there’s also an optional 48dB/octave Low Cut filter at 20Hz, which can be useful, because the LFE section is capable of producing extreme low–frequency signals.

The centre section is called Output Levels, and provides output-level faders and monitoring control for final level manipulation of the unfolded audio. The channel format selector switches between SMPTE, DTS and Film Standard formats, affecting the routing of the audio through the plug-in and determining both the input and output channel format. Each output channel has a delay line, level control, solo button and mute button. The L/R and Ls/Rs pairs have addition link buttons. A further button allows you to check how the output sounds when ‘folded down’ to stereo, while you can also bypass the unwrap process altogether, which gives you the chance to compare the original signal with the processed signal. TC Electronic Unwrap.

• TC Electronic Unwrap

This plug-in is a software version of the Unwrap algorithm from TC Electronic’s System 6000 hardware processor, which measures phase, delay and spectral differences between a pair of stereo channels and uses the results to create a 5.1 mix. The interface has TC’s familiar structure, with a Main page and five sub-pages, though the presets provided on this plug-in are excellent; I found all of them ‘just worked’ and so hardly needed to go into the various sub–pages to make adjustments.

The Main page allows you to trim the left and right input signals, either to balance an off-centre mix or to bring the overall level down and stop Unwrap from overloading. The L/R Processing parameter determines how much the L and R front channels are processed. At zero percent, Unwrap only adds sound to the four other channels, preserving the original left and right outputs as they were, while a setting of between 60 and 70 percent will usually preserve the width of the original mix even after a Centre channel has been added. The Main page also lets you derive an LFE signal from the input, using a low-pass filter.

The first sub-page deals with the Centre channel and allows you to shape the frequency content of the Centre signal. The Ref Level control should be set to match the typical input level; you can then choose between ‘Contour Styles’, and finally apply some EQ to the Centre channel if desired.

On the page that controls the Surround channels there are Decorrelation, EQ and Contour controls. Again, you set the Ref Level, then choose between the Contour Styles, and select a Decorrelation style that complements your program material. Make sure you try the different Styles, as they suit different types of material, but try them out with the Focus control set at zero. Once you are happy with a Style, increase the Focus control to taste. This is all easier to judge if you solo the surround channels. Then finally adjust the Decorrelation Tone and EQ parameters. There’s also a Downmix page, allowing the user to check for stereo compatibility; you can set how much audio to extract from Left/Right, Centre, LFE and Surround channels.

• SRS Circle Surround Decoder (Xtract)

We’ve already looked at SRS Labs’ Encoder and Decoder in our round-up of Dolby and DTS encoding plug–ins (June 2007: www.soundonsound.com/sos/jun07/articles/ptencoders.htm), but now SRS have redesigned their Decoder so it can be used for more than just monitoring. With the new Xtract function, mono, stereo and legacy LCRS content can be unwrapped into a surround environment for delivery or mixing in either SRS Circle Surround or discrete 5.1. There are three Xtract modes, which can be used to obtain a surround sound image as a base for repurposing old stereo and mono content. There are no included presets with this plug-in, but with relatively few controls I don’t think that’s a great problem.

SRS Circle Surround Decoder.When you click on the Xtract button, the Content Type section changes to offer three options. SRS do not describe what is actually going on in each algorithm, so I can only explain what I found trying them out. The ‘2-Channel Cinema’ algorithm is designed for converting stereo film and television content to surround and produces a wide immersion effect, which would not leave enough at the front for some styles of music, but is great for effects. The ‘2-Channel Music’ algorithm produces a much more front-heavy output, with mainly reverb and ambience in the surround channels, and would suit conventional music performances where the music is coming form the front and the room is behind. Finally, the Mono algorithm is for converting mono film and music content to surround, and produces a pleasant enveloping output especially when you consider that it’s coming from a mono source.

In addition, SRS’s Post Processing options work in Xtract mode. SRS TruBass is a bass–enhancement algorithm that uses proprietary psychoacoustic techniques and can be applied either to the LFE or the main L and R front speakers, while SRS Dialog Clarity is a patented algorithm apparently designed to improve dialogue intelligibility in the source material. These Post Processing options really come into their own in the Xtract mode, and both do what they say — almost too well, so ‘little is good’ is definitely the order of the day here. I found I didn’t need to go beyond three (out of 10) on either to get a very impressive effect.

One bug with this plug-in is that the Pro Tools plug-in Bypass button has no effect. The workaround I found is to select Monitor in the Process Mode section and set the monitoring to Stereo, which will pass the input signal through unprocessed, thus providing a bypass path. Once set to Stereo, it is simply necessary to switch between Xtract for the processed version and Monitor for the unprocessed version. This plug-in really needs output solo and mute buttons so you can hear what is going on in the various channels.

Test Material

I selected a range of test material to try out the different types of application you might want to use an ‘unwrap’ plug-in for, with the aim of testing particular issues that can cause problems with these processes:

• Mono male speech over stereo piano: this sample also has reverb on the speech, so will the reverb on the speech end up in the surround channels?
• Mono female speech over stereo piano: this sample is very good at tripping up time– and pitch-correction software.
• Orchestral mixed track: what happens to the strings? Do they end up sounding as though they are playing under water?
• Contemporary mixed track: does this produce a convincing sound?
• Sound-effect wind: does the wind sound envelop the listener, or not?
• Sound-effect birds: again, does it envelop the listener, or do we simply end up with a mirror image in the surround speakers?
• Sound-effect traffic: does the traffic end up passing from front to back, or both in front and behind at the same time?

The results are summarised in the information below. The point of including sound effects was to see how well an ‘unwrap’ plug-in would handle using stereo sound effects in a surround production. If we can’t get this to work well, it could be a costly exercise to replace all our stereo effects with surround versions — if they even exist!

Conclusions

• Waves 360 Surround Imager

Until now, this was the plug-in I used to unwrap stereo material in a surround mixing environment in Pro Tools, and I have been very satisfied with the results. However, this plug-in was definitely the least effective of the ones tested here — although it could be argued that it is a clever surround pan pot, and so it is possibly being pushed beyond its intended use here.

• SRS Circle Surround Decoder

This was the unexpected star of the bunch in terms of what it was able to achieve, with the possible exception of image stability on the mono speech samples. However, to be really useful, it does need some of the features found in the other ‘unwrap’ plug-ins, such as output solo and mute buttons. The bypass bug needs sorting too. The Post Processing options are an added bonus for beefing up less-than-good content, and if you need to encode into a matrix format as well, this plug-in is excellent value for money.

• Cycling 74 Upmix

From my initial experiments, I thought I was going to be disappointed with this plug-in. However, I was always able to find a preset and setting that worked well with each sample. I would prefer that the Pluggo interface were dropped and it became a proper RTAS plug-in, as the need to create a separate track to feed it is a real pain.

• TC Electronic Unwrap

This plug-in handled all the samples very well, and has a good selection of presets. The good news for anyone who already has a System 6000 with Unwrap presets saved is that you can import them into the Pro Tools plug-in too. This plug-in always could be set up to be faithful to the original sound and then, if desired, set to enhance it too. This plug-in comes top of the class, but for me, the surprising thing was that it only just made it.