Delay Widener next steps - mid/side components

  • ckemper Carrying on a bit of our discussion on Delay Widener (I didn't want to 'derail' the acoustic thread and figure this makes sense to be here)

    I'm interested in your thoughts on further development of the Delay Widener. I've looked at the settings I use on a digital live mixer to achieve this process; I retain a regular mono channel for the the centre component, with the signal duplicated into both sides of separate stereo channel, with L and R delay set at 7.5ms and 10ms respectively so that both are separate in time from the centre component but not so much that the delay is perceived, creating a faux-mid/side workflow. They can be faded in to adjust the 'width' and a high pass filter set to 150Hz with a comparatively gentle 12dB/octave slope value prevents the earlier L component skewing the stereo image left, especially with the unaffected centre component still present and upfront.

    The separation of mid and side components in the Delay Widener and the ability to add a basic HPF to a 'sides' component only would be a great enhancement in my opinion, reducing the number of channels/DSP paths needed on a live desk if people wanted to achieve this. It would also simplify the process greatly, as the live engineer would not need to craft it during setup, or during the first few gigs of a tour, it would be done at source. This opens the effect up to analogue desks or smaller digital systems as well, which rarely seem to have a bank of outboard delays or other dedicated processors available for this stuff.

    Too niche, or potential next steps?

  • You hit a sore point. Here are some general words about the Delay Widener.

    I have published the Delay Widener quite late in time, only by the frequent request of users.

    Still I am aware that the side effects of using delayed channels are not well understood by the majority.

    When being used in recordings, it sounds nice, nearly as a double tracking. But you lose the mono compatibility and risk comp filter when mixed down to mono.

    And we have users critically noticing the sound leaning towards the non-delayed sound. But this is the Haas effect that they are creating.

    In a live venue some FOH mixers do not take into acount the different positions of the audience, in relation to the PA. Once you walk away a few meters from the centerline only, you lose the phase coherence and have a natural "Delay Widener" effect, that you cannot avoid.

    You are aware of all this.

    My question is however: how would you benefit from a non-delayed center channel, making the signal drifting back to the middle (Haas effect) for the audience positioned around the center line. And how would the audience positioned to the left and right benefit from mostly listening to the delayed side channels?

  • I'm aware my suggested method is a bit of a hack as it is for a single audio source rather than a true mid/side mic technique using a fig-8, and side channels need to be used sparingly as a result. It's a shared workflow with some other live engineers though, so I'll try and explain as best I can.

    How would you benefit from a non-delayed center channel, making the signal drifting back to the middle (Haas effect) for the audience positioned around the center line?

    I've tried to retain the mid/side workflow at least by using the centre 'dry' channel for the following reasons:

    • To provide the full frequency range of the source without HPF - HPF is used to reduce the Haas effect for side channels.
    • For use as the primary point source in the mix (so some pan could apply within the stereo mix), then act as a reference around which the sides are constructed/faded in, always retaining a point of reference. The idea is that the Haas effect of the L channel arriving first will have been reduced enough by the HPF that it won't 'disturb' the pan position of the main component as such. (The placement of L and R components can also be done relatively if the main component is panned, as you would a stereo reverb, if you use a stereo bus send set to 'follow channel pan').
    • To fall back on exclusively if the mix destination is purely mono (PA or broadcast), or if comb-filtering effects in a particular venue are noticeable enough to render the effect unsuccessful. You can just mute the sides channel rather than having to change delay values.
    • If there is a guitar solo, the centre channel can be pushed to bring focus to the centre, just as the guitarist struts to the middle of the stage for a while. This is easier than boost + pan operations, requiring one fader only, and you know it'll work for the mono destinations too.
    • For compatibility with, centre fills, mono repeaters (possibly LCR PA systems), which are likely to use a mono downmix of a stereo master via a matrix rather than multiple masters with 2.0, 3.0 and 1.0 formats. Although the HPF reduces the noticeable effects of comb filtering in low frequencies, a mono downmix of the sides alone will have lost those low frequencies entirely; the dry channel as the main component provides them.

    And how would the audience positioned to the left and right benefit from mostly listening to the delayed side channels?

    The dry channel is still the most prominent component of the sound that they hear; the delayed channel is faded in only enough to generate the effect. I referred to the dry signal as 'centre' but if it needed to itself be panned it could, as described above. For the moment, let's say it is still located centre. Audience members left of centre would therefore perceive the dry signal to their left, as the amplitude is greater from that speaker and it arrives first. The L delay signal would come exclusively from the L PA out, with the intention of expanding the image between C and L, so that wide image of the guitar moves with them relatively the more left of centre they are. There would still be some pickup still from the R PA side and mono front fills that contain the other stereo components, downmixed. The interactions are surely only as complex as if you try to mix mono through multiple PA zones, but again if it's not working the sides channel can just be discounted. I think the point at which this method comes unstuck is when audience members are very close to a singular PA point source. If it's a centre fill it's ok as the dry component takes precedent and mono compatibility is there for this purpose. If it is either the L or R speakers exclusively, then there's not much helping them any case, as they can only ever be listening in mono, but at worst they will be listening to C + delayed L, which I suspect is only as complex in terms of phase interaction as the reflections coming acoustically from the venue.

    Now, what we really need is a PA tour technician that has to figure out these interactions with a SMART meter all the time, as I could be way off the marque in some of my claims without realising it.

    If were to simplify things for a moment though, do you think just a subtle HPF on the left channel of the Stereo Widener might reduce the Haas effect enough to please those that notice it?

  • You are the guitarist of your band?

    Have you tried such an installation before, or is it just theory?

    Why do you plan such installation just for the guitar amp, and not for the other instruments?

    Wouldn't it make sense to include other instruments into this big picture?

    Why do you think a HPF at 150 Hz would have an impact on the Haas effect?

  • Yep, a guitarist as a hobby - I don't play on a scale that would warrant a large format digital console capable of these processes, unless I steal one from work and hire a van to shift it (it has happened). For a day job I'm a Systems & Support Engineer for a console manufacturer. I count myself fortunate to work with a lot of live, broadcast and studio engineers internationally on tours, festivals, concert series or other live-to-air events. Primarily I work through the preparatory stages of facility commissioning, tour rehearsals/first gigs, provide operator training, often through first shows or station broadcasts as well as offline events. It's the broadcast work that has drilled mono-compatibility into me, but live pop/rock/world music tours where stereo widening techniques seem to be deployed readily to, at least in part, clear up busy centre-focused mixes and make a narrow stereo mix seem wider than it is without hard panning.

    I don't see the workflow as just for guitar; it's usable for any single point source sound, and yes the context of an overall mix would be considered. It's also born out of practical use and experimentation rather than just theoretical; I've either tinkered with it myself when mixing or worked through it with others who are far more talented than I, putting it to use it on backing vocals, small brass sections, widening mono effects returns, a host of world music instruments or in some cases anything in a busy centre-focused mix that can afford to be lost in the mono down mix due to the cancellation. This is where the centre channel can help - there aren't many guitarists that would agree they are sonically dispensable. Typically, it's the sort of workflow question that emerges when a tour has gotten underway and engineers look for subtle refinements they can add in now that they are recalling the mix settings from previous nights and not having to start from scratch each time.

    Having settled on this mono+sides workflow when required, I'm still trying to work out the best way to combat the more obvious remnants of the Haas effect, and have been trying out all-pass filter effects to rotate phase of particular frequency bands on just the early channel. I got poor results when trialled at a tour rehearsal though as it seemed to exacerbate comb-filtering effects. An engineer at a theatre I regularly visit mentioned he used the HPF trick on the early signal to try and improve things. This is different to 'usual' widening workflows I've seen, where a low pass filter is used on the delayed signal to keep the effect without adding to the 3k and above busy part of the mix. I'm not convinced of that method, but it seems reasonable though that if you high pass the earlier of the two delayed components (left through this example) you can reduce the chance of comb filtering with the centre component or the right channel, and the inter-aural amplitude difference favouring the R component will slightly offset the effects of inter-aural time difference favouring the left. To my ears, this is what seems to happen. These later conclusions are indeed more 'theoretical' explanations to what I perceive, so it would be good to get someone else's opinion. Worth experimenting with though. I'm by no means an authority on the subject of widening though, having 'found my way in the dark' to get to this point, with some experiments and collaboration. This is why I mentioned a PA systems tour tech or a seasoned FoH engineer would be a good person to comment, given that they are either having to solve phase incoherency or deliberately introduce it for creative effect on a more regular basis than myself.

    It's quite amazing to me what the Kemper can already do. It's this ability to break new ground that makes me keen to throw further ideas into the ring - I shan't be offended if they are deemed as not worth pursuing, just trying to offer something that may be of some use.