<![CDATA[Kyle Hughes Audio - Blog]]>Sat, 06 Feb 2016 00:43:50 -0800Weebly<![CDATA[Tweety Birds and Their Bionic Successors]]>Mon, 30 Dec 2013 18:17:55 GMThttp://www.kylehughesaudio.com/blog/tweety-birds-and-their-bionic-successors
Play the above set of sounds for a fun example of different types of resynthesis.  The first is a recording of a bird in Big Bend National Park, TX.  I've no idea what species of bird it is, but it does have a strange sort of oscillation in its calls.  The spectral view is just as interesting and is included with each clip.  Below is a brief description of each sound.

You know, I seem to say the word "resynthesis" a lot these days (especially here).

Sum of Sines
Achieved by first performing a spectral analysis to determine the sound's fundamental and partial frequency components, then using that information to resynthesize it with (in this case) 28 sine oscillators. Similar to a vocoder, except it uses oscillators in the place of filters.  For this one, I used a Wacom Pen/Tablet to "draw" the pitches by hand, and used pressure to slow time by a factor of 100, effectively "freezing" the sound as it continues to play.

Cloud Bank
Uses analyzed data (fundamental and partial frequency components) to resynthesize the sound through granular synthesis.

Filter Bank
Uses analyzed data (fundamental and partial frequency components) to resynthesize the sound by controlling a bank of filters that open and close to "reveal" pitches.

All of the above methods can be tweaked to the heart's content, sounding radically different when certain parameters are shifted.  For example, you can control the bandwidth of the filters in the Filter Bank, or the number of oscillators used in the Sum of Sines (unlimited).  One of my favorite things about these is actually the spectral view of the cloud bank.

As you can imagine, it took quite a bit of noise removal before I could do the analysis properly.  I am constantly amazed by RX3.

You can see some photos of Big Bend in the next post over, 

New Furniture for the New Year

In other news, I'm building some new furniture for my home studio.  It's a simple 24U rack enclosure, and should get every remaining item off the face of my desk, besides, of course, the keyboard, mouse, etc.  It isn't finished yet, but here are some pics of the progress.
<![CDATA[Field Recording: Summer 2013]]>Sat, 23 Nov 2013 21:47:56 GMThttp://www.kylehughesaudio.com/blog/field-recording-summer-2013 Yes, for the past few months, field recording has been the name of the game.  As previously mentioned, I've been attempting to tackle many themes for The Sound Collectors' Club.  In July, I recorded the Dallas, TX 'city skyline,' as well as a few freight trains passing through Irving and Denton, TX.   The new MKH416 and pair of CM3's have worked beautifully, as have the Sound Devices 702's I'm currently using.   Additionally, I took two separate trips on business to the Texas/Mexico border over the summer, capturing some interesting sounds there.

First up...

Trains: Scarier Up-Close

August 19th, 2013
Looking at the photos, you can see that the train was moving so fast that the iPhone's CMOS camera sensor was too slow to keep up.  This gives it an almost cartoon-like quality, as if it is leaning forward.
There are two categories of trains I went out to record.  The main one was Trains: Design, while the other is for background train textures.  When thinking of a location to record trains, Denton was the first place that came to mind.  There are trains all over Dallas, but I can recall a number of times recording in Denton when trains showed up unexpectedly, making their presence unmistakably known.

In particular, there is a bridge with two sets of tracks on it; as far apart as the tracks may be at the other end of town, they are but a few yards from each other here, and must pass over this narrow point.  Sure enough, within about six or seven minutes of parking my car, I scrambled to ready the gear as a train blasted its horn.  Just in time, I hit record and plugged everything in.  The first thing rolling was the R26, directly underneath the train, with both an SM57 and AT8004 at its side, in addition to the inbuilt XY and Omni stereo pairs.  Six channels underneath the train, all absolutely terrifying to listen to.

Something to keep in mind with this track is that the left and right channels are recording separate rails of the tracks.  The movement happens overhead, not right to left.  Also, the microphones do have different pickup patterns (cardioid and omni).  I chose to bring two different mics so that I could hear how each of them sounded separately.
Next up, I plugged in the a pair of Line Audio CM3's configured in ORTF.  They sat inside a zeppelin on a stand.  This track gave me a great sounding pass-by.  The credit for this goes to Rene Coronado for blogging about it first (recording a train, no less).

Finally, the MKH416 sat in my hands as the train blew by.  There was actually quite a strong wind being pushed along when it arrived, as well.  The sound is wonderful, but it's in mono, and just doesn't have that "larger than life" stereo image of the ORTF technique.
Since it was over 100 degrees out (I know it doesn't look that hot in the photos, but believe me... it was), I went ahead and moved on to another position by the time the next one came.  I moved under the bridge and eventually got a few more over the next few hours.   I couldn't find any information on schedules for these freight trains, so my best best was to just show up and wait.

I had to leave some of the microphones unattended while I set the others, and the poor 416 just kept getting the short end of the stick.  The cable I used for it after I climbed down into the culvert cut out halfway through one of the bys, and the recorder was out of reach on one (It had started raining, so I left the recorder under the bridge with the other mics, and pistol-gripped the 416 at the edge).  One time, I left the mic perched atop the edge of the culvert, around 40 yards from the tracks.  When I climbed down to monitor the CM3s, the 416 got blasted so hard with the horn that it clipped beyond saving.  I know this sounds like bad practice for a sound designer, but I really had no idea how it was going to sound whenever I moved.  Hey, lesson learned.  Even the R26's internal mics clipped when it sat underneath the train.   Believe it or not, gain staging wasn't the problem- the digital tracks themselves did not clip- not even close.  The mics themselves couldn't handle the sheer SPL of the train at times.  Fortunately, all is not lost, and the sound is still quite usable.

Personally, I think the real heroes here were the CM3's.  We'll see, though, next time I go out.  This isn't the first time I've recorded a train, but it is the most thorough I've been capturing one.  I'm thinking next time I'll use two AT8004's; I prefer it over the SM57 for the tracks-track (heh).
<![CDATA[Multichannel Mixing in Reason]]>Wed, 13 Feb 2013 21:57:32 GMThttp://www.kylehughesaudio.com/blog/multichannel-mixing-in-reason For anyone who uses Reason, but always wished for more in the ways of multichannel mixing.

Some time ago I decided I'd like surround panning capability in Reason.  I searched the web, and found this method, which uses the Matrix Sequencer as an X/Y controller in Reason 2.5 (Released May 2003) to achieve 5.1 surround.  This wasn't quite what I was looking for, so I set out to create my own.

Propellerhead's very own James Bernard posted in 2009 about using TouchOSC on an iPhone to send OSC/midi to Reason.  After the iPad's takeover, innumerable apps were developed for this same purpose.  Even Jazzmutant's legendary Lemur lives on via Liine's iOS app.  I stuck with TouchOSC, and this is what I came up with.

Reason Rack Routing

One basic track is comprised of one 14:2 Mixer, and four 6:2 Mixers.  They are tied to the Combinator's Programmer in such a way that the mixers' Pan determines Left/Right, while Volume determines Front/Back placement.  There are independent controls for Center and LFE Channel level.  The buttons on the Programmer can reset the positioning to center (either L/R or F/B), and solo the center and LFE channels.  Overall track volume can be scaled with the first and only fader used in each track's 14:2 Mixer.  Or, if you route the Direct Out from each track to the input of its own 14:2 Mixer, you can use the Master Mixer's faders instead.

For the template, I've set up four Mix Channels and four audio tracks.

All the individual tracks are summed through four more tracks containing 14:2 Mixers to form each discrete channel.

TouchOSC Templates

In TouchOSC, eight pages bear an identical layout, but send different OSC and MIDI values, and are thus assignable as eight separate but equal track controllers.

In Reason, the quickest way to assign software switches or faders to hardware is to enter Remote Override Edit Mode.  In TouchOSC, two faders beside each X/Y Pad exist for this purpose.  One sends the same message as the X-axis on the pad, the other duplicates the Y-axis.
The other two faders and four switches correspond to the remaining Combinator controls for each track.  The solo buttons are toggles; the pan resets are momentary.

Also included are a few more custom TouchOSC layouts.  One page in particular is mapped to the Reason transport.  In any case, it's likely you'll have to assign each parameter manually once Reason is open, as well as provide output routing specific to your equipment.

The master volume for all channels is in the Master Section Insert FX.  Because of the unusual setup, the master fader does nothing, nor does the master bus compressor.  Reroute if you dare!

Hopefully this helps anyone wanting and waiting for a decent surround panner in Reason.  Feel free to post any improved or modified versions below.

The Goods

Snatch the download here or on the Downloads page.
Reason Surround Mixing (ZIP)
File Size: 131 kb
File Type: zip
Download File

<![CDATA[The Importance of Meaningful Art]]>Wed, 13 Feb 2013 02:15:46 GMThttp://www.kylehughesaudio.com/blog/the-importance-of-meaningful-artWhile browsing Facebook, a friend posted this question:
When viewing 98% of Art that I'm told is important, what I'm thinking is, "Why does this exist and why do other people care about it so much, and why am I learning so much about it!?!?!?!"

I just wanna make video games.
One thought lead to another, and after a few very long replies, I decided to post it all here, instead.  My response follows.

Important art is made with intent; especially those made in a time characterized by the lack of such motivation.  Evoking feelings and making statements through visual medium is the art of it.  The information is presented to you all at once, but the story of its creation and author keep you thinking, unraveling the mystery one minute after the other, piece by piece.

What determines a piece's importance? Consider this:

"It is the glory of God to conceal a matter;
But the glory of kings is to search out a matter."
Proverbs 25:2

That said, one of my long-time goals has been sound design for video games.  The thought of bringing that essence from still and motion picture's long history into interactive media moves me toward that goal.  Also, it sounds like a ton of fun. ;)

There's a reason why Hollywood content is marketed at big-money target audiences.  For as much as people notice the similarity in protagonist "heroes," they fail to realize the marketability of a tall, young, white male who "gets the pretty girl," or a movie targeted at African Americans; ie, Tyler Perry and "Black" movies.

What's more, a great number of video games aim for what I would call a more extreme and focused target audience.  The mark of a popular video game is often one that supplies constant stimulation.  Consequently, developers are creating single-serving games that contain nothing but guns-blazing action, while still others create genre games of their own- all in the name of sales.

Games give the illusion of the player controlling the content, when in reality, the player only reaches as much as is allowed at any point.  This leaves room for discovery, which sends a much more profound message than it being served, in a similar manner to the way on-screen action in a film trumps a direct explanation or narration via dialogue.  This discovery can be used to the storyteller's advantage, as a 60 hour time investment can pan out and pay off much more beautifully and fully than a two hour film.  Game developers simply have the time that directors do not.  They can tell the story that painters cannot.  This, of course, is the polar opposite of the extreme action games- but, the two can exist as one.

Braid's creator knew this.  Look into Bungie's Halo universe (not the one perpetuated by 343 Industries).  The Biblical representation is astonishing, and quite obvious at times.  The hero- Master Chief, sacrifices himself, though he lives.  The unmerciful Flood, the old Covenant, and even the number seven- Bungie didn't get those ideas from nowhere.  And then, the not-so-obvious; Cortana and the related Durandal from Marathon are significant in their virtual (see: spiritual) representation as well.  Remember The Legend of Zelda: Ocarina of Time and its Seven Sages that guide the pure-hearted Hero of Time?  I could go on, and perhaps I shall, should anyone ask for more.

I'll conclude with a link to this article by Keith Stuart, titled Are Video Games Art: The Debate That Shouldn't Be, and a friendly reminder that video games, open-endedness and all, are fundamentally different from all the other art forms I mentioned.  Even the optimistic comparison to the early days of film should be stopped before it starts.  The real question is, how do you make the most of what you have?

<![CDATA[Thunderstorm Synthesis: The Tempest]]>Fri, 08 Feb 2013 19:09:39 GMThttp://www.kylehughesaudio.com/blog/tempestI've been posting up a short recording of a swelling storm labeled "Thunderstorm Synthesis."  What this actually is, and how it's created is still a little convoluted, so I think it's time I gave an explanation.
Here's the description I've posted on SoundCloud:
The genesis of this track is an attempt to synthesize and sync a realistic sounding thunderstorm to picture. This recording consists of 4 samples of rain, and another 3 samples of rain+thunder that I recorded one afternoon.

Equipment used was the inbuilt mics on a Roland R-26, and a Sennheiser ME66 into a Sound Devices 702. The clips were recorded at 96kHz/24-bit, and they were processed at 48kHz/24-bit.

For processing, I put the samples into Kyma, and crossfaded for texture.

The howling wind sound is an analog-style low pass filter's frequency, level, and resonance being controlled by a Wacom Intuos4 pen/tablet.

The rain slowly swells, which was done by changing parameters of a granular reverb.

The thunder was also controlled by the Wacom tablet, with X, Y, and Z (pressure) dimensions mapped to making the thunder swell in level, density, and texture.

This could have been output in surround, but I don't have that many monitors ;). This style of "rain-synthesis" can also go on indefinitely.

Let me be clear and admit that saying "synthesis" doesn't mean that what you're hearing is entirely synthetic.  As mentioned above, I did record samples of rain; about five minutes of a light shower, but I'm only using around 30 seconds of it.

The real goal is to turn that 30 seconds into a high resolution storm- that doesn't sound like it's been looped.  This isn't a place where I want to bend a metal sheet or use a sizzling sausage to add tone to a film (those methods, mind you, are not beneath me).

Another part of this is to do an exercise in Kyma.  If you're unfamiliar (most are- I ask everyone), Kyma is a niche sound design environment by Symbolic Sound Corporation.

Setting the Pieces

There are other ways to achieve what I'm attempting, but for my purposes, I record four channels from Kyma into Pro Tools.  I could go up to eight channels, but the hardware I have keeps me at four.  Truthfully, I don't even have four monitors, but I tell Kyma the intended "virtual" speaker placement, and it will output a quadrophonic storm regardless.  In the event that I need only a stereo signal, the ambient nature of this allows me to simply trash the rear channels, rather than downmixing.
Kyma Speaker Placement
Click images for a larger view

After recording, the samples should be cut and faded into loops.  The fades can be indiscernibly short, if desired.  This isn't essential to making the process work, but does create smoother transitions, eliminating obtrusive "pops" when each file cycles back to its start.

Since I'm dealing with such short clips, this is a good idea.

Beginning in Kyma with the signal flow, you can see the process starts with a just four samples.  Suffice it to say, the better your source material, the better the end result.

There are two amplitude crossfades for texture, and optionally, the wind and rain can be controlled at the same time.
The "wind" is simply a low-pass filter with four poles.  The gain, cutoff frequency, and feedback can be changed as its source resonates through it.  At the input of the filter are the mixed rain samples; since the frequency range of rainfall is so broad (almost white noise), it keeps a constant level around the cutoff frequency of the filter.

In this view, a Sound begins with its source on the left, and travels to the right.

It's then written to and pulled from memory via a sample cloud.  What this means is that it slices the audio into grains, or small samples, of a specified length, and applies a set number of other processes before playing the sample back, all while continuously reading what has just been written from the source itself.  Yes, it seems this is a form of granular synthesis.  In this case, however, the grains can be much longer than classic granular synthesis typically allows.  The short grain duration was mainly due to a lack of processing horsepower, but that was some time ago.  Instead of grains that are 1-100 milliseconds long, these samples can be as long as the user wants.  In my case, I've set the grain duration to about 5 seconds.

If that's a little confusing, imagine this:

You have in front of you a book full of blank pages.  A man approaches and begins writing.  Page by page, he writes.  He does not slow down, take breaks, look over what's written, or even look up at you- he only writes.  A woman approaches and starts to read what the man has written.  She deliberately stays three pages behind him.  She reads consistently, and at the exact same pace as the man.  Except that as she reads, she cuts the pages into small pieces of confetti and blows them into the air.  Quite remarkably, it is a continuous process.  Her eyes remain fixed on the pages, but her hands and lungs work to keep a constant, linear flow.  You can vaguely see what's written on these small strips of paper, but you perceive it as a whole, more than any individual word.

Her name is granular synthesis.

You grow tired of this nonsense, and ask the woman to make the strips of paper bigger, so that you can read them as they blow by.  She asks, "Which size?"  You answer, "Umm, I don't know- six square inches."  She immediately adjusts and cuts the pieces differently.  They now float down in pieces of two by three inches.  You can not only read words, but phrases, and even sentences.

This is fascinating.  Why is he still writing?  Doesn't he notice her destroying his work?  But again, your excitement fades.  You want something more.  You give the man sentimental pat on the back for holding up like a champ, then you ask the woman for more.  You don't know how, but just more.  She complies.  With a free hand, she manages to make copies of every page and cut unique strips (still six square inches) from each one.  Astonished, you give her further instruction to improvise.  Again, she asks for boundaries, which you provide.

Ok, enough of that.  The sample cloud can read and duplicate audio, as well as apply a set amount of random, yet repeatable, deviation ("jitter") to each parameter.  One of these parameters is the pan, which is one reason I chose to output in four discrete channels.  Each new sample played from the cloud can be in a different place spatially.  When many samples are triggered together, the effect is considerably immersive.
Memory Writer
Memory Writer
Sample Cloud
Sample Cloud

Watching Them Fall

When the Sound is played, we see the Virtual Control Surface display all the specified hot values.  By changing these, we can effectively determine the storm's intensity.  Direct controls the amount of direct signal sent to the output, while Reverb attenuates the sample cloud.  Density is an interesting variable, as it limits the number of grains being played.  Here, a low Density value sounds mellow and sparse.  When setting Density to maximum, the Sound begins to take on the natural chaos of a storm.  When the other values are driven upward, the volume scales appropriately, so that the great amount of beating raindrops are multiplied in both power and number. 

There are other values displayed here, like Source, which determines the input gain, before it reaches the Memory Writer.  WindLevel, WindFreqhz, and WindResonance control the level, frequency, and feedback of a howling wind, respectively.  I originally included this with the rain in order to record both simultaneously, but later separated the wind into another Sound.
Light Shower
A Light Shower
Rumbling Storm
A Rumbling Storm
To manipulate the VCS on-the-fly, I use a Wacom Intuos4 Tablet and Pen.  Kyma recognizes the tablet right away, and allows for several dimensions of control.  
For example, to create a wind track, I'll assign WindLevel to the Y-axis, WindFreqhz to the X-axis, and WindResonance to pressure (Z-axis).  By slowly moving up and down, the level rises and falls, while simultaneously drawing the pitch and depth of its "howl."
Intuos4 Pen Parameters
Intuos4 Settings in Kyma
Cracking Thunder 3 VCS
The patch for thunder utilizes sub-bass, which operates almost identically to the wind, but resonates at much lower frequencies.  Remember that nonsense about random deviation?  Here, anything with "Jitter" attached sets the fence for how much its affected function is spread.

So, if PanJitter is set to 1.0, the samples from the cloud can jump around anywhere from Left (0.0) to Right (1.0).  If I change it to 0.5, then they can land anywhere from Center-Left (0.25) to Center-Right (0.75).

In addition, a low pass filter can be activated by bringing the cutoff down into an audible frequency range.  It sits at 24,000 Hz in this image because it was initially implemented to cut out most of the rain in the recordings.  After I obtained clean (rain-less) samples from The Hollywood Edge, I left the filter in, but changed its cutoff to the Nyquist frequency as a bypass (48 kHz / 2 = 24 kHz).

The Tempest

Finally, it's time to record.  All that's necessary at this point is to arm the tracks in Pro Tools and begin "painting in" the rain.

I've recorded another example, to illustrate how simple it can be to work with multichannel audio in Reaper.
Pro Tools Rain Result
Pro Tools
Reaper Rain Result

After some quick editing and mixing, we are finished!  For this, I used Avid's built-in compression plugin, as well as Waves' MaxxBass for enhancing the lows and low-mids of the thunder.  Check out this screenshot of the edited and mixed product, then scroll down to hear how it all comes together.  The exact example you're viewing is "Thunderstorm Synthesis.A2"

Please note that SoundCloud compresses anything and everything it hosts, meaning a broadband ambient track like this won't sound great unless you hear the uncompressed original.  To download the uncompressed file, click the "download" button in the top right corner of the SoundCloud player below.

For more Signal Flow variations, click the numbers to the left of the image below.

Lastly, you can download this encoded AC3 file to hear the quadrophonic storm on any consumer surround system.

The Tempest.ac3
File Size: 5083 kb
File Type: ac3
Download File

As I record new samples, create new examples, and discover new techniques, I'll update this post.  If you'd like to discuss this or have comments, corrections, etc., leave a comment below or send email to Kyle@Kylehughesaudio.com.

<![CDATA[High SNR for Extreme Dynamic Range Audio]]>Mon, 31 Dec 2012 17:56:00 GMThttp://www.kylehughesaudio.com/blog/high-snr-for-extreme-dynamic-range-audioFor a while, I've been following a recommendation for field recording; it suggests using an XLR Y-cable to record two tracks from the same source.  In my case, this source is usually a shotgun microphone at the end of a boom pole.  It may seem redundant (it is), but over time, I've found it to be a method worthy of mention.

I touched on this over on the Documents page.  If you're interested, go check out the downloadable sound report templates.  I find this to be particularly useful in recording location audio for film, but it also shines when close-micing foley sound, or just about anything else at the right place and time.

First, understand this: options are good.  It is certainly better to have extra microphones and channels, should you need them.  However, there are situations where this isn't an option.  For example, you have a limited selection of microphones, channels to record to, or limited space to record in.  Recording dialogue in the corner of a cramped room (next to lights, camera, other crew, etc) doesn't exactly cater to having two boom operators or setting up mic stands.

Why, though, would you need two tracks for a single source?  Well, when your subject goes from a whisper to a shout, and back again, the ground begins to shrink beneath you.  Barring asking the director or talent to change their delivery, your choices are few:
  • Ride the gain, and compensate for the shifting noise floor in post-production
  • Use a limiter on capture, while the signal pumps and shifts automatically
  • Capture the scream on location, while the whispers risk remaining in the mud
  • Bite the bullet- mix for the lower signal, and ADR the scream.

Of course, this "scream" could be a loud response on impact, unpredictable ambient noise, or anything else that clips too far to be reasonably captured on the same track as a low level signal.  Also, ADR and foley may be out, perhaps if you're recording an interview or documentary.

So, on we go, in search of a solution.

And here, we arrive, at splitting one source to multiple channels.

Female XLR to 2x Male XLR
From the beginning to the end, the signal travels from the microphone, through a single cable, split through the Y-cable, and into two independent channels, each with it's own preamp and gain level.  I have found the most painless procedure to be recording an "A-track," mixed as if it were the only one, while a "safety" track with a lower gain awaits something louder.  This means that editing is the same as always, but that no content is lost, and can be recalled at a slightly lower signal-to-noise ratio (SNR) when needed.

In fact, Propellerhead's own Balance interface has a feature called "clip safe" that performs this very function when used with Reason.
"Clip Safe is a magical lifeline for distorted takes – a red-eye tool for recorded audio. Clip Safe records your take on two channels, one at a lower volume. If your recording clips, it lets you substitute it for the other one. Don't ask us how it works, it just does."
The main drawback to this technique is that if you need it, it can involve heavy editing and some significant processing.  For sources that frequently jump from barely audible to barely tolerable, the time required in post can be bothersome, at best.  This is a sacrifice, but for mission critical work, time in post is insignificant compared to losing a recording of a rare event.

Moving forward, the tracks are imported to a DAW or NLE of choice, and editing begins.  I'll be using Pro Tools 10.  I'll also scale the waveform view to show detail; in other words, the level of each track is relative to the other, not to the size of the waveform depicted.  The "flat" part is still 0 dB.

Important:  If you are first synchronizing sound to picture and editing visually, make sure not to delete any audio tracks, even if they clip!  It will sound better, but it's best left to audio post.  Of course, if you're aware of this process, you can edit at your own discretion.

In this first screenshot, Tracks 1 and 2 are shown, recorded with the method described above.  It isn't anything special, and is simply a verbal slate, followed by a clap and a cough that clip Track 1.  This will do for a demonstration.  Click the images for a better view.
Source Material
The flat lines at the top of the waveform signify overmodulation of the signal. This is bad! Track 2 does not clip, but the verbal slate before the clap is significantly softer than the same portion of Track 1.
Editing in this manner isn't too difficult; I've chosen to select only the distorted parts of the clip, rather than the entire impulse.  This is possible because the content is exactly the same on both tracks.  I find it easier to make small cuts and fades here instead of treating it as if the tracks were mutually exclusive; any noise brought up will hopefully be masked by the actual signal.
Selection Made
After the selection is made, it's separated, as is the safety track.
At this point, the distorted audio is removed; however, as long as the selection remains unchanged, it will audibly jump down, then up by the difference in gain initially set on the recorder.  To rectify this, we can use clip gain to quickly even it out.  Alternatively, it can be done using the Audiosuite Gain plugin.

This is where quality preamps are important.  I've raised the gain on this clip by 12 dB, but the difference can easily be much higher.  The higher the difference here, the more your noise floor will come up.  Fortunately, if your B-track has a strong, clean reading, the noise shouldn't be a problem.
At this point, we can trash the backup track, since this is the last cut being made for this example.  Two small fades (the minimum in Pro Tools is four samples long), and we're ready to mix and process.

But beware!  If we consolidate it into a new file now, we could lose the information over 0 dB that we worked so hard to save!  A floating point bit depth would allow us to retain it for normalization later on, but instead, we'll group the clips to process them altogether.
Mix and Process First!
Above, the early consolidation leaves us with the same problem we started with. Below, the signal hits far higher than 0 dB, and will clip the D/A converters if left unchanged.
Now, it's time to process.  We can treat this as any other sample.  We can do this non-destructively and in real time with an insert or send (the group has its own clip gain, and can be ungrouped at any time), or use an Audiosuite plugin to process and consolidate, all at once.
With a little compression, the VU meter on the left no longer peaks.
So, what are we left with?  The dynamic range we've been able to capture is considerably large, and the noise floor is nice and low.  Using a low-budget solution (under $20), we've saved ourselves a lot of hassle (or being hassled!) rescuing or recreating damaged content.

Other possibilities exist, such as implementing a multi-band limiter in field recorders, though this kind of processing would likely require more than a simple firmware update for existing equipment.  Maybe it's already around.  I don't know, but it's definitely something I'd like to see.   As far as front end processing goes, you can mix and process before the signal hits the recorder, but it's costly to obtain the necessary equipment, takes up precious space, and results are rarely perfect.

Zaxcom's NeverClip has my attention.  It uses two Analog to Digital Converters (ADC), operating as a function of the recorder, rather than manually routing through two preamps.  I'd like to try it for myself sometime.  Check out this demo.
Anyone with thoughts on this is welcome to comment.  You can reach me via email at Kyle@kylehughesaudio.com for a discussion over this or a request for a post on anything in particular.