Quote:
Originally Posted by satshanti /img/forum/go_quote.gif
It is my claim that by using the VI settings of my latest post, hardly any information present in the stereo mix is lost or hardly any additional distortion is introduced, but through the smart use of a complex mixture of cross-feed and phase shifts, the spatial cues that are available in the stereo mix are re-created binaurally. VI uses ambisonics for that, and Dolby Headphone is a really good and precise algorithm. It does work with general HRTF only, so I assume the algorithm might have a different effect on different people with different ear sizes.
I'll give you an example of how stereo spatial information is transferred into a binaural space. Imagine a vocal track. Now, I'm not a sound-engineer, so I don't know how all these effects are called, but I imagine that a recording studio is not a bathroom, so any reverb,delay and/or echo effects one hears on a vocal track is added "artificially" via a DSP afterwards. This effect is then merely "placed" somewhere in the stereo mix. VI recognizes this and places it in the virtual space where it should be namely all around you, as if you were in a room with the singer in the middle. And this happens for all instruments. With most recordings this whole process creates a very realistic and natural-sounding sound stage, similar to a real binaural recording, meaning a few things are placed even behind you, which is something to get used to. The process does not add delay AFAIK (although through VI one can enhance certain effects).
Anyway, I personally prefer to listen to all of my music like this, and I do try without it once in a while, but quickly put it back on.
|
Mmmm, it's not quite that simple (never is!). Using your example of a vocal track: Depending on the type of vocal it may be recorded with the real reverb of a venue or studio, as good quality commercial studios will have well designed acoustics in the live room. In this case an array of mics (including stereo mic'ing) may be used to capture both the direct and reflected sound. Positioning is critical though as it's next to impossible to totally eliminate phase artifacts. It's just as likely though that the vocal in popular music has been recorded in a dry acoustic and then reverb added artificially. Most reverb processors are quite sophisticated. Let's say you pan the vocal 75% to the right speaker, you can also send the vocal to a reverb unit also 75% to the right. The stereo reverb unit will then calculate reverb within this stereo space, in our example the Early Reflections (one of the many parameters of digital reverb) would be generated first in the right output of the reverb unit before the left output. This is to maintain the perception that the relative positioning of the vocal track in the stereo soundfield is re-enforced by the acoustic information produced by the reverb unit. Reverb units (outboard or plugin software) are complicated bits of kit and very DSP heavy as they ideally needs to be able to process the audio sent to them as many as 3,000 times a second and each one of these 3,000 reflections needs processing variations to introduce a certain amount of randomisation and avoid correlation (which causes nasty artifacts, like ring modulation). The quality and realism of reverb units vary enormously and you can tell this from the price. Reverbs vary in price from free downloads up to about US$14,000 for a top of the line model. Although a top of the line model will be more complex still, as it will calculate the timings and positions of the reflections for a full surround soundfield (as well as a stereo soundfield when set to stereo mode), these units tend to have a whole array of DSP chips to handle the complex algorithms. What is more, it's not unusual to use two or even three different reverb units in a single stereo mix (to create varying depths).
As a general rule though, naturally recorded reverb usually (though not always) gives the best results, although it requires a lot more time, effort, skill and expensively designed live room. It's not uncommon to mix both live acoustic reverb and digital reverb.
It can get even more complicated than I have explained but I hope I've demonstrated that the process is far more complex than just adding an effect and then placing that effect somewhere. The interaction of reflections from digital and natural reverb in your average mix can be incredibly complex, very unpredictable and often requires considerable skill to create depth, while not loosing clarity or separation. There is no way to calculate the effect of all these interacting reflections, it just requires experience, taste and a good set of ears (and obviously a good monitoring environment). Although reverb is based on science, it's use and application is an art. Also, although a stereo mix can be analysed, it can't be separated out to it's original constituents and re-mixed by a DSP process, taking into account all the reverb interactions.
All the above leads me to feel very dubious that a stereo mix can be re-processed adequately with the added timing (phase) information which will be introduced through the use of crossfeeding and compensations for HRTF. There is just too much unpredictability and complexity for there to be any way to guarantee that this added phase information is not going interact in some way with the original mix (phase cancellations and/or inaccurate positioning). This unpredictability of what is likely to happen is why I stated that for the purist it's probably not a good idea. Although depending on; the mix, the effect of different playback equipment and each individual's subjectivity and physical attributes, some people may prefer the effect and some may not. Some people may prefer it only for certain mixes or genres and on certain equipment.
I am used to critically analysing mixes in my head and listening for the use of time based effects (Delay, Echo, Phase and Reverb). So I personally have never heard a crossfeed or DSP processing consumer system which I liked or felt comfortable with. I think in general most audio engineers and producers would feel the same way. It does concern me that a consumer is adding processing to a mix which may change or even nullify some of the artistic endeavour and hard work I've put into creating it. I like to think though that I've come to terms with the fact that once a consumer has bought my product it's theirs to do with as they please (except ripping it off of course) and that it's no longer "my baby". I also realise that listening to music with cans or IEMs is already a compromise, so I can appreciate why efforts are being made to modify the output of cans to emulate a more realistic listening experience.
Sorry this is such a long post, it's not easy to explain without being able to actually demonstrate with a mix and production system in person. Hopefully though it's all useful background information for those of you interested in the technicalities of how music and audio products are created.
G