Audio |
|
Auralchemy How to Get Stereophonic Gold from Monophonic Lead (1)
What It's All About
In theory, I find this delightful and enlightening. In practice, it can be daunting and exasperating but the results speak volumes.
It the case of this article, to finish the analogy, the single channel of audio from an amplifier is any monophonic recording, the designated frequency range is that of a particular sound source within the recording and the designated speaker in a cluster is the potential location of that sound in a stereo field. That makes you the crossover and despite the illustration, you're an active one. There are many possible uses for this technique throughout the vastness of all possible audio production. I'll be illustrating a simple one with a real world example close to my heart using Sound Forge or Cool Edit Pro, Acid, a few DirectX plug-ins and a fairly simple series of routine parameter adjustments.
What is mono? Mono, or monophonic, here refers to an audio format in sound recording where discreet channels are mixed together into a single track then played through a single speaker. Think of the Victrola with its disk and horn. Please! The ubiquity of digital audio means future generations won't likely know of to appreciate the existence of analog audio technology, or the insightful experiments it can contribute to. Few serious audio systems these days rely on a single speaker so, other than from a phone, an answering machine, an AM radio or a portable TV, the monophonic experience you're most likely familiar with is the playing of a single track through two speakers. (2) If you want a memorable example, listen to any old Robert Johnson recording and marvel at how much performance could be captured by such primitive technology. The mono of importance here is that of a single track of recorded sound in a single data file that can be made into virtual copies and EQed separately in Acid. Mono Case Study: Bomb @ Cocodrie, San Francisco 9/11 1999 In my real world example, I have video documentation
Neither one of us got great sound (Hugh, bless his heart, didn't even shoot the entire set) and I feared the worse until it occurred to me how to use a computer to get the most I could out of what sound we did manage to record. While I ultimately used this technique on both mono tracks to produce an exponentially better stereo mix, the process I'll be describing is for a single cameras monophonic sound. I mention now that this technique can be performed on a number of different mono recordings of the same sound just to tantalize you; in case you happen to have more than one sound recording device laying around, gathering dust. (4) Obviously, depending on the source, different mono tracks will have different sonic profiles and each will offer unique characteristics and challenges requiring varying degrees of attention. That attention will be spent, in large part, defining the actual and ideal sonic profiles of each sound (or instrument) of interest, noting bandwidths where sounds of interest crossover, or overlap, and ultimately using graphic EQs to emphasize their strengths, de-emphasize their weaknesses and remove unrelated noises as much as possible. (5)
In the case of Bomb 9/11 1999, I was dealing with the sonic profiles of an amplified tenor voice, electric bass, a small but dynamic drum kit and two wildly F/Xed guitars. Sounds naturally overlapped each other across the spectrum so I was resigned that anything resembling true instrument isolation was unattainable and that the best I could hope for was spreading targeted bandwidths across a stereo space to differentiate them. For aid in deciding which bandwidths to target, I created charts of each instruments ideal and actual ranges then began prioritizing their likely contributions to either the clarity or energy of the performance. I wouldn't begin to decide placing them in a stereo space until I actually heard my results. This is, after all, simply a visual aid and not an alternative to listening. Look At What Mono Has To Offer There are a variety of approaches to visualization available to you that can aid in determining what the characteristics of sound are, including the audio visualization hack I wrote for #X "Quick and Dirty Frequency Analysis." It describes a process of creating a chart that identifies the characteristic range of an instrument and then comparing it to a chart of the actual recording of that instrument, characteristic range being a mythical ideal and recording being, at best, an approximation of the past. Regarding the example I am about to present, it supposes that I have already followed steps outlined further below (and elsewhere in this book) to produce charts for the sonic profiles of both the ideal range and the actual recorded range of the bass. Assuming you have, at least, access to a passing familiarity with hack #X "Quick and Dirty Frequency Analysis", I'll forego describing how I created these charts and get straight to analyzing them.
After a number of decisions, selective targeting of the latter frequency range (8) produced the improved chart on the right which shows that the lower bass tones, with the energy, are now present enough to be heard roughly equally as the higher bass tones, with the clarity. Another pass with the EQ to remove as many of the other instruments as tolerance allows and then it's on to the next instrument. Once all of the instruments are charted, and significant bandwidth conflicts noted, it's time to excavate the sounds. Open a file in Acid and get ready to make even more decisions. Extracting Instruments In the simplest terms, after you've made a two track version of your mono recording, you could EQ your left and right channels differently by putting all the bright sounds this-a-way with a high-pass filter and all the bass sounds that-a-way with a low-pass filter. Don't expect much. As long as the sound is comprised of a number of discreet sources, arbitrary decisions where to boost what will as likely exaggerate noise as diminish it. Under these circumstances, it's inescapable that identifying and controlling bandwidth overlap between instruments is going to be Sisyphusian Task Number One for the duration so it's necessary here to create some reasonable expectations for success from a set of charts sufficiently representing the details of the discreet sources and their overlap. The comparison charts above illustrate the determination of a targeted range (the ideal of a particular musical instrument; E2 - G4 @ -3dB), what I have to work with (the actual sound as seen by Windows Media Player) & what might be done about it (using DirectX plug-ins). At this point, it's the ideal of the instrument minus the realities of the track that determines the number of copies to make and how to prioritize them. The larger the number of slices you choose to divide the available spectrum into is bound to produce more detailed and refined results requiring more detailed and refined attention. Is it worth it? It's really a matter of resolution and the energy you have to spend pursuing it. In the end your results are all that matter. Besides, we never know what we're missing, do we? Certainly not enough to be disappointed by expectation. How much effort is really worthwhile? As in all things that may steal your time, consider the source, and be assured perfection is unattainable. As in all things that may test your limits, do the best you can with what you have. Remember, you're making art. Attitude counts. And your real audience wants to give you the benefit of the doubt. Mono Case Study: Continued Anything the soundtrack wants to suggest for a visual editing priority is OK by me. If a mix clues me into something more obvious to emphasize then the less I have to rationalize every single editing decision. Did I just say that out loud? Will MTV kill me for spilling the beans? Before I begin editing the video, I've tirelessly listened to the original mono Bomb and I'm really hoping to produce five unique tracks for eventual mix in a stereo space. I'll happily take what I can get.
Since, in this recording, the vocals and the cleaner parts of the drums are in the same range, it might be best to work on them together leaving three tracks. Without a doubt, I'm keeping the guitars, the bass and the vocals apart for discreet treatment later. That I happen to have another mono recording from the show may double the discreet tracks in the end, making six, more or less. Whatever the number of final tracks, they represent targeted ranges so let's go about determining those first. For current purposes, the ranges that interest me are: (9)
Upon a casual perusal, I would guess that the range where the bass and guitars overlap (approximately 164.81 Hz to 261.63 Hz or E3 - C4) will be a useful reinforcement and not necessarily a problem spot, though I may eventually want to separate them in stereo space.
Assume I assumed the position and made charts based on my three ideal bandwidths from the table above and the original track. The bass chart above, for instance. I've determined the problem spots and I'm itching to get mixing. I've opened the track in Acid, assigned a 20 band graphic EQ, copied the track three times and reduced the frequencies in each tracks' EQ that fall beyond their unique targets. Parametric EQs can also be useful in isolating the instruments of a track. I've picked graphic EQs simply because they offer predefined frequency ranges and curves. Paragraphic EQ are the best of both. EQ Multiple Mono Tracks This is what it looks like in Acid.
In this example, I've boosted the target bandwidth of the bass by +3dB and reduced the overall gain of the track by -3dB, effectively leaving the target where it was while reducing frequencies out of range. This approach may not appeal to you. As in all things, do what you think is best if it works. Now it's time to do and notice. I'll start by soloing the bass track and adjusting different EQ sliders, looking for the telltale signs of either clarity or energy. In time, I'll create a copy of the bass track. One will be EQed for clarity and the other for energy. To Bounce Or Not To Bounce Bouncing comes, or is that goes, in two directions. Neither one of them up. You can bounce an isolated group of instrument tracks down to a single instrument submix for later mixing with other submixes or you can bounce all of your tracks at the same time into a final mix. The advantage of the former bounce is that it may make a complex mix more manageable. The latter bounce offers expedience, probably, but it depends entirely on the specific complexity. Don't we all. In praise of simplicity, three tracks immediately suggests I do hard pans like voice\left bass|center guitar/right, or one of the other combinations of three tracks and three pan positions. In this case, I'd rather not do the obvious so I'm not going to commit any track to a place in stereo space yet. Besides, I have enough PC power to easily preview a mix of 32 tracks, each with multiple real time DirectX effects being applied. These days, it doesn't take much to work marvels. As more tracks and pan positions enter the mix, the complexity grows approximately exponentially. Pace yourself and don't commit to decisions before you have to. Many of the decisions I make can be made at different times in the process, depending on what information I need in order to make other informed decisions and when information appears to make them. Bounce Multiple Mono Tracks If I'm going to bounce multiple tracks to a single track, I will always bounce to a stereo track so I can try to maintain some control over each instruments contribution of energy and clarity to the whole. Per instrument, I typically use a channel for each. Even in this simple way, I effectively have six channels to mix from; Bass Left = Clarity, Bass Right = Energy, and so on. Mixing Sets of Multiple Mono Tracks By superimposing my instrument ranges over the EQ, as in the example below, it's apparent to me that I should start by keeping the clarity of the bass and the energy of the tenor as far apart in space as possible. The same for the clarity of the tenor and the energy of the guitars.
In Acid, I add three busses, copy each instrument as needed and assign all copies of a particular instrument to a unique bus. These groupings create three instrument submix master faders. Up the screen, on the individual tracks, I assign Pan positions that should provide a fairly wide distribution and press play. So begins the creative decision making process.
This is one suggestion for a mix session starting position. I can not count how many more there are, much less where they could lead, but I know this configuration will quickly tell me if what I hope to get out of my source is possible. It will also aid me in deciding if the effort is worth it but only in an advisory capacity. Actual investigation is the only measure of truth and the truth should dictate the decision. And by all means, copy complete instrument groups, mute the original, and try another EQ and Pan scheme on the duplicate. If you assign that duplicate group to its' own bus, it's quite easy to toggle between Group busses during play back for comparison. If that's not enough flexibility for you, you're not trying very hard to complicate your life. Final Thoughts I hope you've enjoyed this cautionary tale, and that you'll allow me one more analogy before I go. As as audio enthusiast, I liken the removal of noise, or rather the excavation of sounds and their placement in space, to the art of sculpting where the removal of material reveals one of the countless latent shapes within any block of stone. The metaphor is strained but intact. I suppose that makes a computer a chisel to a slab of mono sound. Hmm. Guess what the hammer is. You may commence banging your head. You're welcome. -Michael Joseph Woody Addendum Here is the 2 camera video I used to illustrate this tutorial plus an edited version showcasing an improvement in audio utilizing the techniques described above.
Footnotes 1- This is a companion piece to A VisuAlchemical Silk Purse: How To Extract Stereoscopic Views From Video, hopefully found elsewhere on this website. [return] 2- While the same audio information is suppose to be coming through both speakers, we none the less hear both speakers distinctly, having two ears in two different places. But that's about the most difference we ever hear between our ears, unless one's blocked or overwhelmed by competing sound. Which is great because that's how we can locate and differentiate the sources of sounds in space without seeing them. If there aren't too many sources or reflecting surfaces. [return] 3- It was the 2nd to the last Bomb performance, as of this writing. Not long after, Michael Wareham Dean went south and became the DIY legend he always knew he was meant to be, in part by performing for the public, making ground breaking films, launching the $30 DIY School series of books for Muska&Lipman Publishing and repeatedly dragging it all around the globe to show you and your cousin how to do your own versions of it yourself. On behalf of appreciative fans around the world, thank you, Mr. Dean. [return] 4- Examples of common devices that record audio and tend to collect dust include: Computers, VCRs, Video Cameras, PDAs, Cell Phones, Answering Machines, Boom Boxes, Cassette Decks, Dictaphones, Wax Cylinders, etc. The list does go on. Each one has its charms and warts. In conjunction, as in a mix, the total may be greater than the sum of their parts. [return] 5- Which is pretty much the reason why graphic equalization exists in the first place, I reckon. I don't pretend to be shedding any light on the technology itself but rather I'm pointing your attention towards what may be an other-than obvious feature of it. That requires stating the obvious from time to time, the need for which may not be obvious, which is why I mention it now. Obviously. [return] 6- A bass guitar is typically tuned low to high (looking down if your holding it right) E2, A2, D3 and G3. G4 is the highest sting played on the 12th fret, at the octave, which most basses possess. I tend to initially concentrate my EQ on the top half of the neck from the nut, E2 to C4, and then venture out as needed. [return] 7- This chart is a fanciful composite created from the bass guitar over time. I use it in order to increase the probability that all the notes in the song, that can be heard, are represented in the chart. [return] 8- The ranges you set are determined by the sounds you are working with and the energy vs. clarity problems they uniquely pose. FYI: Information about the ranges of a wide variety of musical instruments & more, is available on the internet, for the inquiring mind. [return] 9- I have to confess that this example is a bit of a fraud. For one thing, I'm not sure Michael Dean is technically a tenor and for another, the range for the guitar represents its tuning and frets but doesn't take feedback and distortion into account, which can be and was a major factor in the startling sound of Bomb going off. I expect due consideration will be given for the spirit of the thought, if not its strict adherence to the truth. [return]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
With a fairly common computer and relatively inexpensive software, any monophonic recording can be duplicated, each copy can be EQed to target the characteristic bandwidths of specific sound sources (instruments) found within and all can be remixed discreetly in a stereophonic field to further define and differentiate them. Even relatively strong monophonic recordings can be rejuvenated with this technique. Further, two monophonic recordings of the same event can produce exponentially better results.
I shot the show from the front of the stage to get performance highlights and I had my friend Hugh Kinniburgh (of cable accesses' Reality Check TV fame) shoot from the back of the crowd to get the overview. In effect, I have two mono recordings of the same show from different places in the club.
Differentiating Jay Crawfords' and Doug Hilsingers' guitars may be a lost cause so it might end up only being four tracks. But don't bet on it.
The voice lands well within the range of the guitars so that overlap will definitely need problem solving attention. It's a good thing there aren't any woodwinds. Or a piano, which can require attention from 27.5 Hz to 3520 Hz, and beyond. Well beyond once you consider harmonic resonance.
As they too are all over the spectrum, in small pockets of percussive tone fields, I'm giving it up that I don't have much control over where the drums will end up popping through in the final mix. I hope to push the tuned instruments into the foreground if I'm lucky, vocals first. I doubt Tony Fags' rhythms will be hard to miss, wherever he ends up.
