Creative Audio Processing
01 - Realistic Recordings

 

 

First, we'll study a case in which there is no style whatsoever, and in which one must not be creative at all. Here its is : We have to record a piece of classical music, and we are supposed to get to a result which is as realistic as possible. To achieve realism, we're going to consider that we must use the purest, more pristine way to record the piece of music. Basically, we'll record in a nice concert hall, and we'll use only a pair of DPA electrostatic microphones (formerly B&K). There will be no post-production, no EQing of any kind. Here, creativity is some kind of an anathema. Sound is only here to serve music, everything must be absolutely transparent. What's on the record should be as close as possible to what's in the recording room, whatever that may mean. This method is actually used by many classical music engineers, and is presented as the ultimate method to record classical music at the Conservatoire de Paris. It is somehow an "old guard" method. It is an elegant method. But, most of the times, it fails. This method is a mistake. Now, many things can be learnt while trying to understand why it doesn't work.

 

Four reasons (why it fails)   1. Let's say that I was recording a symphonic orchestra. Do I want to have a realistic sound in the first place ? The answer is no : the record will likely to be played in someone's living room, and nobody can or even wishes to have a symphonic orchestra in their living room ! Why : first, the orchestra's dynamic range is simply too great. If I manage to cram a whole symphonic orchestra in my living room, I'm lucky if my neighbors will do nothing worse than calling the police. Furthermore, recreating a symphonic orchestra with a pair of speakers is just not possible : even with the levels right, the original orchestra's size and width can never be recreated, certainly not with a pair of speakers in a living room. So what's the point of using only two microphones, of sacrificing so many possibilities, just to pursue a goal that's just unreachable and then fail ? 2. Now let's say that I was recording a piano piece. I wanted to be as transparent and "realistic" as possible, so I put my pair of DPA4010 at the room's critical point, I adjust everything so it sounds like a real piano in my studio environment, I don't do any post production... And then the music is published, people download it from the ITunes music store, they listen to it with their IPod, and it sounds all shitty. In fact it's barely audible. So much for my desire of realism. I wanted a "real life" piano, but I got shit. That's because I had forgotten that in real life, a piano piece is not played through IPod headphones ! Now if I really want to get a realistic piano sound through IPod headphones, I will have to forget about any kind of "accurate" recording. I will have to cheat to get that piano to sound right through IPod headphones, and that's all. 3. I still have my piano recording. I've got it to sound as realistic as possible through IPod headphones. But then, people listen to it on their stereo, in their living room. Will the recording still sound realistic ? No, not really. In my headphones, I had the piano plus the surrounding acoustics. On speakers, I have the piano plus the surrounding acoustics, plus the living room acoustics. Roughly, it means that people will listen to something similar to what they would hear with the real piano, should they be not in the room in which the piano plays, but in the room next door. ... and nobody wants to listen to a piano from the adjacent room, even with the door opened ! So once again, my desire for accuracy, for realism, was a mistake. I might as well forget about realism, and instead I should try to make my record sound generally good, whatever that means, and good on as many listening systems and situations as possible. Really, it seems that the record has a life of its own - it's quite resilient to any try to get it to sound "real", so I decide to forget about that "realism" issue. 4. ... so I get back to basics, and I want to record a saxophone, and I just want the sound to be nice. But then I ask myself, what is the sound of a saxophone in the first place ? Is it a saxophone from the saxophonist's point of view ? No, that would probably be too close. When listening to a saxophone, people generally don't hug the saxophonist. Then is it from the audience point of view in a large concert room ? But then from what row ? The front ? The middle ? The back ? How large the concert room should ideally be anyway ? Then it occurs to me that some studio sound engineers put microphones on stage, not too far from the player. So is that from where I should ideally listen to a saxophone ? Because if that's the case, it sounds quite differently than from the audience. Then again, I've seen other engineers recording a saxophone by putting their mikes like 20 centimeters from the player, and that was in a studio. Does that give a more accurate saxophone sound ? It shouldn't : people never listen to a saxophone this close. Now I'm confused. .. the answer to this quandary is simple : there is no accurate recording of a saxophone. There is a style of music, a style of player, a style of recording. All those should go together. The important thing is that the result sounds reasonably good on many systems, while being adequate to the music.

 

All those examples lead to an important fact : a recording is always a fake. To give a good impression of something that you want to capture on a record, don't try to be true, know how to do a good fake. A good example : in sound design, to get a good impression of a stream, one never records a stream : streams are "created" in bathrooms, not recorded in the wilderness ! All stream sounds in the movies are fake. More generally : being closer "conceptually" to the original sound scene (few mikes, no post prod....), that doesn't mean being closer to a realistic result. Result is the only important thing, and any methods that can get to good results are valid. When I press "play", if it sounds good, it's good... no matter the method that' used.