For News Editors: we’re looking for content aggregators who will keep up on interesting articles, editorials and library releases; providing “block quotes” from said content and links out to our friends and fellow conspirators in the community…perhaps with the occasional nugget that appeals to the gear lust in all of us.

For Contributing Editors: we’re looking for more people to do what we already do. News, reviews, interviews and insights into the many sorts of multimedia audio we take part in.  Contributing Editors will be expected to do much more than a News Editor and long interviews, editorials and articles which will require more effort and time.

Requirements:

-Reliability.  You will be expected to make multiple posts in a month (for News Editors in a week), sometimes at short notice. And you will be expected to do so *every* month. If you are incapable of this please do not apply.

-Autonomy. We are looking for self-motivated and self-sufficient audio professionals who also want to help out the audio community without being having to be poked to get work done.

How to apply:

-Send your portfolio (Audio and/or Writing) as well as a writing sample to: volunteer@designingsound.org

-State in the email subject which you are applying for: “Contributing Editor” or “News Editor”

If you do not have a writing sample then please make one before applying. The writing sample can be an example news post, a full article or a review of a product.

Applications will remain open for the entirety of May 2013 and we will begin selection in June.

“Where are we?”

“Connecticut.”

“No. How far are we?”

“From what?”

“Yesterday.”

Directed by Sean Durkin, Martha Marcy May Marlene (2011) centers around the damaged psyche of Martha, portrayed by Elizabeth Olsen. Martha is a young woman who, through the time spent living as part of a small farm commune, has lost nearly all sense of boundaries…from social through temporal. I believe “nearly” is the appropriate term, because she leaves the commune in response to certain events. If she had lost all sense of boundaries, that probably would not have happened. Despite retaining this small level of faculty, Martha is lost. The five lines above this paragraph are a conversation between Martha and her sister, Lucy, from early in the film. In my opinion, this is possibly the most illuminating exchange that occurs in the entire piece. Martha wants a description of the distance she’s traveled from a temporal reference point. She spends the film slipping in and out of time and place, pulling the viewer into her fractured perspective of the world.

These time slips are made all the more effective through Coll Anderson’s sound design. The visual edit by Zachary Stuart-Pontier makes use of human perceptual idiosyncrasies to establish an opportunity, but it is the use of sound that draws the viewer into Martha’s flow. In her world, time is not linear…nor is it circular. For Martha, time simply is. The past and the present are concurrent, a state which increases her confusion, fear and paranoia as the film progresses. Let’s take a look at some of examples of how sonic elements help control the viewer’s perception of time.

Our first example picks up where the quoted conversation above leaves off. We hear the work of a person with a hammer a short distance off-screen…perfectly natural for the scene we are watching, and continues through the shot change. The next shot, however, does not take place in Connecticut; where she is staying with her sister. Martha is back on the farm in New York, sitting in the grass and holding a conversation with some of the other residents.

The temporal displacement is repeated at the end of this exchange. We hear a male voice from outside of the frame, “I’m going in. You wanna join?”

Martha responds in this scene from the past (New York), “Sure.”

She stands up, arriving back in the present (Connecticutt). It’s then that we realize the voice she just responded to actually belongs to Ted, her brother-in-law. The visual transition…her beginning to stand up in the past, and finishing in the present…takes advantage of the fact that we tend to accept the temporal continuity of a matching cut on action. The present and the past have blended momentarily. While changes in the visual elements are immediately apparent, sound has the ability to covertly occupy both temporal locations. This sequence and its use of sound is the first indication we have that Martha is trapped between both timelines.

There is another scene, later in the film, that employs a similar approach. Present and past are both taking place in the kitchen of the respective houses. In the present, Martha is assisting her sister Lucy with a meal. In the past, Marcy May (Martha’s name on the farm) is assisting Katie. While preparing the food in the past, Martha sneaks a piece of food and is scolded.

Martha eventually responds, “Sorry Katie.”

The scene holds in this kitchen from the past as we hear Lucy ask, “Who’s Katie?”

Cut to the present, where Martha picks up again, “What?”

It’s interesting to note that, in order to make this exchange between Martha and Lucy work, the film goes in and out of ADR six times here. A lot of careful construction was required for a relatively brief moment in the film. It’s a very important moment though, as it’s one of the key scenes that brings Martha’s fractured perception of reality to the forefront. Sound in this film is a purely subjective element, filtered through the lens of Martha’s psyche. We hear what Martha hears, and that helps cement the viewer in her perceptual world.

But is what she hears real?

There’s a sonic element in the film that appears three times in total; twice in the “present” and once in the “past.” I’m presenting these in quotations, because of the ambiguity that’s inherent when you consider that we’re hearing only what Martha does. Martha hears the sound of something hitting the roof at night. This sound could be considered foreshadow or flashback. It is both and neither, because we are never given confirmation that the sonic event is actually occurring in the present. Is it another time slip that Martha is only experiencing sonically? It could be, as Lucy suggests, only pine cones falling on the roof. The fact that we can’t answer the question ties the viewer into Martha’s increasing state of confusion.

As the film progresses, these carefully constructed time slips reveal more of the events that caused her to flee the farm. The fear and doubt she experiences in the past are mirrored by her increasing anxiety and paranoia in the present. There is a parallel escalation between the two time periods…and “parallel” may not even be the correct description, if we’re holding to the idea that she experiences them concurrently. These time slips would not be as effective in portraying this without the sonic world of one spilling into the other around the transition points. They are crafted to draw attention to themselves, for both Martha and the viewer, but only to do so after the moment has passed. The revelation of the illusion gives it greater weight.

I spoke to Coll Anderson on the phone before posting this article. Everything you’ve read so far is my personal interpretation of the film, uncolored by the conversation with him (excluding the anecdote regarding ADR in the mentioned kitchen scene, of course). I wanted to first confirm that my interpretations were correct and the intended aim of the film’s structure. They were. I also had the suspicion that this use of sound around the time slip transitions/visual edit points was planned from early in the production process. Coll caught me completely “off guard”…it wasn’t. This particular approach to structure was something that Durkin, Pontier and he developed entirely in post-production. Don’t misunderstand, the overall concept of her distorted perception was in place prior to production. The exact implementation, the use of sound versus visuals, is something they developed through experimentation and collaboration. They took the time to explore the material, and crafted something haunting and beautiful out of it. That fact is a credit to Durkin’s creativity and flexibility.

I’ve intentionally ignored several other scenes and sonic elements that contribute to the flow and perception of Martha’s time in this film. I chose to do so in the hope that you will watch the film yourself. There are some truly subtle and clever uses of sound and dialog in this film, and I would not want you to watch it without experiencing at least some of them in a natural state. Check out the film. Find the other points were sound pulls you into Martha’s head. Then come back and share your thoughts in the comments below.

I’m on a plane from San Francisco to New York, May 3, and I see that the Designing Sound theme for May is time.  Two notions relating to time and sound design come to mind.  One is that fiddling with time in storytelling is always a playground for sound.  Jumps to the past or the future in a movie swing the door wide open to using sound in a subjective way, and sound is almost always most powerful in a story when it expresses or reflects subjectivity.  Our ears are subjective tools.  They are easily tickled and excited by ambiguity.  In the final version of Coppola’s rewrite of the John Milius script for Apocalypse Now the first line of description is: “Coconut trees being viewed through the veil of time or a dream.”  As he made the movie Coppola gradually turned Milius’ brilliant, but mostly objective, script into a carnival of subjectivity.  We hear and see the war through the highly filtered senses of those young American soldiers.  Walter Murch was one of the chief designers of the filters, and I had the incredible luck to be there as it happened.

The second idea that comes to me about time and sound design has to do with the stretch of time from pre-production to post production on any given project.  It makes no sense that sound design is relegated to “post production.”  I think of the wonderful scene in the Coen brothers’ film Barton Fink when the title character checks into the hotel.  He walks across an empty, smoky lobby to the clerk’s counter, but no clerk is there.  He taps the bell once, but the bell rings, and rings, and rings, and rings way longer than a bell like that should ring after being tapped only once.  When it has been ringing for about twenty five seconds the clerk finally appears, climbing up through a trap door on the floor behind the counter.  He calmly touches the bell with his finger to stop the ringing.

Obviously, the Directors wanted to make this place, this hotel lobby, feel odd.  They had a variety of tools at their disposal to accomplish that goal.  Many less imaginative directors would have chosen to use an odd piece of musical score as their main sonic tool.  The Coens chose to use no score at all.  Instead, they staged the entire scene around a bit of sound design.  The strangely lengthy bell ring, and the odd twist it gives to the place, are the focus of the sequence.  The actors, the cameras, the lights, the props, etc. were directed to serve the performance of that bell.  Why aren’t more scenes structured in this way, allowing a sound idea to influence the creative decisions in all the other crafts?  One reason is that we, and the directors we work for, are used to thinking of sound as icing on the cake.  The last thing in the process.  The best sound is baked into the cake beginning in pre-production, from the earliest…. possible…. time.

Speaking of time, as I mentioned before, I’m writing this on May 3. Regarding tomorrow…  May the Fourth Be With You.

The Recordist

Machine Guns HD Pro SFX Trailer from Frank Bry on Vimeo.

Presenting Machine Guns HD Pro Sound Effects Library. This 24-Bit/96kHz collection features four fully automatic weapons recorded at a wide open 400 acre ranch in North Idaho. With a mountain ridge in front and open field behind, the guns generated a breathtaking echo while at the same time the “close up and personal” microphones capture the intense muzzle blast and mechanics of each weapon in stunning detail.

SoundMorph

SoundMorph is a new addition to the Independent SFX crowd. They’ve currently got three libraries available (Robotic Lifeforms, Users of Tomorrow and Road Riders) with a fourth one on the way. Of particular not is their “Users of Tomorrow” library, a library of “user interface” elements which includes a custom built software synthesizer that will allow you to generate even more custom sound effects in the tone of the library. Check it out:

Rabbit Ears Audio

This Library is all about cold! The temperatures of these files range from a low of −20°F up to a balmy 35°F. Our Rabbit Ears froze so your ears would not have to.

Winter Atmospheres was culled from some of the coldest forests and wilderness areas in North America. Our frozen locations include: Nemadji State Forest and Superior National Forest in Minnesota, Crex Meadows Wildlife Refuge in Wisconsin, and Algonquin Park in Canada.

Soundeffects.ch

The Room Tones Sound Effects Library contains 130 room tones, including 12 rooms presented for the first time in 11.0 Auro 3D sound. All sounds were originally recorded in 5.1 surround sound or 11.0 3D sound and are available as stereo versions.

Room Tones Version 1.0 includes electrical room tones; room sounds with different ventilation noises; room tones with clocks; quiet rooms; specific room sounds etc. All sounds are sorted and categorized.

The Room Tones V01 sound effects library not only includes over 130 different room tones, it also contains the IR1 Impulse Response Set

The IR1 Impulse Response Set contains 54 sets of multi-channel impulse responses, some of them from the same locations as the Room Tone sounds. The impulse responses are intended to be used in creating multi-channel room tone atmospheres. Hence they were recorded in multi-channel format up to 11.0 Auro 3D sound. All impulse responses are available in mono, stereo, quad, 5 channel and 11.0 Auro 3D sound. All available configurations are accurately described and documented.

Boom Library

Get realistic outdoor reverbs – fast and easy – with BOOM Library OUTDOOR IMPULSE RESPONSES

Using impulse responses, you can apply the acoustic characteristics of a particular location to your target audio as if it had been recorded in that place. With our BOOM Library OUTDOOR IMPULSE RESPONSES library, you get 68 great outdoor impulse responses to place your sounds into specific locations and make them sound as plausible and realistic as possible. We recorded a lot of different outdoor locations such as fields, forests, hills, mountains, valleys and urban locations. We have ensured that each recording has a rich and detailed reverb tail to capture every characteristic delicacy of the respective location. All files come in high definition audio (192kHz). Each location is documented with a picture to get an impression of the surrounding area. Watch our tutorial video below to see how to work with BOOM Library impulse responses.

Echo Collective

Squelch is a full featured interactive radio instrument and sound effects library.

The Squelch sound library is a collection of 6 radios, walkie talkies, and CBs recorded both through the speakers and in some cases through the line outs. Each device served up a broad palette of static, beeps, clicks, squelches, and random idle chatter.

The Kontakt instrument is a unique performable radio device designed for experimentation and discovery of the broad palette of sounds. It also includes a classic spring reverb unit with a custom interface created from a Pioneer SR202W.

Hiss and a Roar

Tim Prebble is now offering his complete Hiss and a Roar collection as a package deal! That’s 16 libraries containing over 17,000 sounds (92GB worth of sound effects). If you’ve purchased some of his libraries in the past, you can even contact him to get a prorated discount! Check out the full details on the Hiss and a Roar website.

April has ended and May has begun. No matter what we do, or how we may try to control it,
time continues to march forward at its own pace…regardless of our needs or wants. Last month, we had a very specific theme. While there was ample opportunity to explore different approaches and uses, everything focused on a particular type of toolset. We thought it would be interesting to follow it up with a featured topic that is open to a much wider range of interpretations. So, we’ve selected “Time” as our theme for this month.

Sample rate and playback speed, time management skills, pacing and flow, tools and plug-ins that operate in the time domain…these aren’t subjects you would normally associate with one another, but they are all unified when placed under the umbrella of “time.” When you think of the word “time” in relation to our work, what do you immediately think of? And a better question…will you step forward to share that thought here with the rest of the community?

If you’d like to contribute to this (or next month’s) theme, contact shaun [at] designingsound [dot] org. Our theme next month will be “dynamics.”

A hearty “Thank You” goes out to…

• Andreas Jonnson – Designing Ambiences in Max/MSP
• Aurelien Folie and David Sonnenschein – 3 Deaf Mice: A sound-music game
• Tadej Droljc – Sonographic Sound Processing
• Graham Gatheral – A Quick Introduction to SuperCollider
• David Nichols – AudioMotors – Review
• Les Stuck – Designing Sound Discussion Group – Les Stuck on Max/MSP
• Mark Durham – Generating Complexity in Max
• Naila Burney – Using Max/MSP to Build a Fictional Language Dialogue System

Thanks again to all of you. And to the rest of our community, remember that we always welcome guest contributions on the site. If you’d like to share your knowledge with the rest of the community, don’t hesitate to contact us.

…and don’t forget to check out the articles posted by our contributing editors as well, if you missed any.

What is gen~?

Gen~ is an environment within Max that was introduced in version 6.0. It allows for patching of low level operations that are instantly compiled into great sounding objects without the limitations of MSP. I’ve been using it for about a year now and it sounds absolutely amazing. Version 6.1 of Max makes it possible to export gen~ objects as C++ code, making it portable. Cycling ’74 plans to add documentation for VST plugins and the iOS operating system in the near future.

For this tutorial we are going to attempt to create a tremolo AU plugin (sorry Windows users!). We’ll first start with trying to understand how a tremolo plugin works and implement a basic version within Max (I’ll translate object names for the Pd users out there) and then re-create it within gen~ so that we can export the code that will  help create our AU plugin.

DISCLAIMER/WARNING/MESSAGE IN CAPS: gen~ is in beta and there are chances that the code you compile might not work. I’ve got the AU plugin working fine on my computer (OSX 10.7) but haven’t been able to get it to work properly in 10.8. I am also not a coder, I know just enough to hack (or break) my way around things. If you think you’ve come across a bug, do let the kind folks at Cycling ’74 know.

Let us begin!

How does a tremolo effect work?

If you move a fader up and down manually at a fixed rate you would achieve a tremolo effect. Although, its tough to get our hands to move a fader at a fixed rate! If we had to replace the hand with a modulating signal instead that moved the fader, we could create a decent sounding tremolo effect. You can try it yourself. Open your favourite DAW and draw in a repeating pattern on the volume automation lane:
How can we implement this idea into Max (or Pd)? Lets look at some signal processing basics first.

Signals and the -1 to 1 range:

Digital signals are stored as samples that are then read at a fixed rate (the sample rate) to reproduce sound. This is a hugely over simplified explanation of what is quite a complex process. The reproduced ‘sound’ is nothing but a voltage in the analogue domain that is then amplified by an amplifier to drive a speaker. Every device has limitations. A speaker cone moves both forwards and backwards (positively and negatively) and has a physical limitation, beyond which it will not move. Similarly, both digital and analogue mediums have limits. If you try and go beyond the limits you end up with distortion.

If we think of sound as a signal, it is nothing but a stream of numbers that change over time. The limitations of the medium and the system impose boundaries. In digital systems this range is usually between -1 to 1. If a signal crosses the boundaries of  -1 and 1 it will distort (it will not distort within the system, but will distort on output). Here’s a sine wave occupying the full range (-1 and 1), on a digital meter it will peak at 0dbFS

Scaling a signal:

If I want to scale down a number by half I could divide it by 2 or multiply it by 0.5. If I have a number ’10′:

10/2 = 5 or 10 * 0.5 = 5

Or if I have have a number ’1′:

1/2 = 0.5 or 1 * 0.5 = 0.5

If I have signal ranging from -1 to 1 and multiply it by 0.5, I decrease it by half (or drop it by 6dB). Multiplying a signal by a factor allows us to scale it or change its amplitude.
If we could change the amplitude or scale this signal at a constant rate with an oscillating signal, we would end up with a tremolo effect. Simple huh? What generates an oscillating signal? An oscillator of course! The default oscillator in Max (cycle~) and Pd (osc~) spits out an oscillating sine wave in the range of -1 and 1. We could directly patch the oscillator to a [*~] object to scale a signal, but since the oscillating signal ranges from -1 to 1, we will end multiplying the amplitude by a negative number and switching its polarity. Lets scale the oscillating signal to a more useable range of 0 to 1.

We can now use this oscillating signal to modulate the amplitude of a sound:

Quite simple. We can now add more features: make it stereo, add different waveform types, add depth control, etc. Lets first try and implement this basic version in gen~ before we turn into feature creeps.

The guts of gen~

Using the gen~ environment is as simple as creating a new object and typing in ‘gen~’. If you have worked with abstractions and sub-patches this will look familiar. Double clicking the gen~ object opens up a new window which gives you access to the insides of the gen~ object and the gen~ environment.
The default setup includes two inputs being summed into a single output. A point worth mentioning here: everything in gen~ is in the signal domain, there is no messaging system like Max.

Let’s create the first version of our tremolo effect:

First, we need an oscillator. Gen~ by default has a sine wavetable lookup oscillator like MSP, called ‘cycle’ (surprise surprise!). We then need to scale our modulating signal into a useable range (0 to 1) and use it to modulate the amplitude of our input signal.

The patch is very similar to what we had earlier, except we now have an input for our dry signal and an output for the ‘effected’ signal. I have made an additional output to analyse the modulation signal and also included an additional ‘param’ object. The ‘param’ object allows us to use named parameters from the parent patch to control objects within the gen~ environment. The image below should explain this:

This is the parent patch which includes the sound file player, the scope~ object to analyse our modulation signal and the gen~ object that includes the tremolo code. Additionally, we have a number box feeding into a message box named [Frequency $1]. ‘Frequency’ is the named parameter we created in gen~ that is connected to the inlet of the cycle object (to control it’s frequency). The $1 is used in Max and Pd to create a variable value, where the $1 is replaced by any value that is sent through the message object. So, if we send the number ’5′ to the message object in our case, it would spit out ‘Frequency 5′. Or, if we sent it the number ’10′, it would spit out ‘Frequency 10′. This message is sent to the gen~ object which then assigns the number to the cycle object because we have instructed it to do so using the ‘param’ object. Makes sense?

So this all works fine (if you have managed to stay on so far). Lets make this tremolo effect more interesting. I’d like to add dry/wet controls, have two different kinds of waveforms (sine and square) and be able to morph between them, make the effect work with stereo signals and be able to create a phase difference (offset) between the left and right modulation signals.

Lets get cracking!

First, we’re going to get rid of the ‘cycle’ object and instead use a phasor. A phasor is an object that generates steady ramps from 0 to 1 at a fixed frequency (a saw-tooth waveform). We are using the phasor so that we can create both a sine and square wave from the same time reference.

To create a sine wave from a phasor, we need to use a mathematical formula (lots of information about this online):

Output = sin(2*pi*Input)

In gen~ we can implement this using the ‘expr’ object and then scale the signal so that it falls within the 0 to 1 range:
Now, lets create a square wave from this same phasor object. A square wave is nothing but a signal that is at a constant 1 for the first half  of a cycle and switches to a constant -1 for the second half of the cycle. But since we want to control amplitude, we want this range to be between 0 and 1. We can use the ‘>’ object to achieve this. The ‘>’ object is a comparison object that spits out a 1 if it’s input is greater than a specified value or it spits out a 0 if the input is less than the specified value. If we create a ‘>’ object with an argument of 0.5, we can create a square wave from a phasor (it spits out a 0 when the signal from the phasor is less than 0.5 and 1 when the signal is greater than 0.5).

If we crossfade between both signals (sine and square), we can create waveforms that are somewhat in-between both of these shapes. Thankfully there’s an object in gen~ that helps us do this. The ‘mix’ object has three inlets: the first two for the two signals that need to be crossfaded and the third for the control signal. If the control signal is at 0, the ‘mix’ object outputs only the signal it receives on the first inlet. If the control signal is at 1, it outputs only the signal it receives on the second inlet. As you rightly guessed, any value between 0 and 1 is a proportionate mix of both signals. We implement this in gen~ and create a parameter called ‘Shape’ that will allow us to crossfade between both waveform types.

Great! Now lets expand this so that it works for a stereo input by duplicating the patch for a second channel:

 Now, lets use a ‘mix’ object for each channel so that we can crossfade between the dry and wet signals. Lets also create a ‘Mix’ parameter to control the…mix.

Two more steps left. We are almost there! Now that we have a decent tremolo effect working, lets add another paramater that will allow us to offset the left the and right modulation signals so that we can create cool stereo effects. We can achieve this by changing the phase of one of the modulation signals (changing it’s starting point in time). To do this, we will add a value to the output of the phasor and then use a ‘wrap’ object to bring it back into the 0 and 1 range. The wrap objects works like a modulo object – if you are familiar with it. If we give the ‘wrap’ object the arguments ’0 1′, it will make sure that the signal always stays within that range. So, if we send it a 1.1 it will spit out 0.1, or if we sent it 1.7 it will spit out 0.7, or if we send it 2.3 it will spit out – you guessed it – 0.3. Let’s also create a parameter called ‘Phase’ to control the phase difference between both modulation signals and limit its range to 0 and 0.5.

It sounds quite decent already. The only issue I find with it is that when the ‘Shape’ parameter is turned all the way up to 1 (the square wave), it sounds quite clicky. We can fix this by using a low pass filter to smoothen the modulation signal. Unlike MSP, there is no lowpass object in gen~. We need to create one from scratch. Worry not! Thankfully Cycling ’74 have provided an amazing library of gen~ examples (Applications > Max 6. 1 > examples > gen) that include different filters. Lets copy the code for the lowpass filter and set it up so with a ‘Smooth’ parameter, so that when the ‘Smooth’ value is at 0 the filter is at 100Hz and when the ‘Smooth’ value is at 1 the filter is it 10Hz. Lets use the ‘scale’ object for this to save us some of the maths. Also make sure you assign different arguments to each of the history objects (y0 and z0 in this case).

And we are done!

By sending the gen~ object an ‘exportcode’ message, we can get it to export out the object as C++ code. Now lets convert it to

an AU plugin. Crack your knuckles!

I am breaking this section down into a list of steps, so that it is easier to follow (much of below is repeated from the post on the Cycling ’74 Wiki):

1. If you haven’t installed Xcode and downloaded Audio Tools for Xcode, now is the time (check the pre-requisites section at the start of this post to make sure you have done everything right)
2. Download “AUGenExport-v01.zip‎”, from the Cycling ’74 Wiki (available right at the start of the document)
3. Unzip ”AUGenExport-v01.zip‎”
4. Open Terminal (Applications/Utilities/Terminal)
5. If you haven’t used Terminal before, now is a good time not to worry
6. Type “cd ” (without the quote, make sure you add a space after cd)
7. Drag the unzipped AUGenExport-v01 folder from the finder into the Terminal window. This will make the path of the folder show up in Terminal
8. Hit Enter
9. Type: “python duplicate.py MyAU Tremolo yourName” without the quotes. Replace yourName with your name or your company name or whatever. This is a python script that will duplicate the example Xcode project provided by Cycling 74 and rename it to the plugin name we have specified (Tremolo) and the your name or company name
10. Hit Enter
11. The python script should do its thing and create a folder named ‘Tremolo’. If you don’t see it you have done something wrong (obviously!)
12. Go to the Max Tremolo project and send an “exportcode” message to the gen~ object
13. Choose the new Xcode project folder we created (Tremolo) when prompted and hit OK
14. Navigate to our Xcode project folder (Tremolo) and open up the Xcode project: Tremolo.xcodeproj
15. Click on ‘Tremolo.r’ in the Project Navigator column on the left
16. Change value for COMP_MANUF from FRED to your company name
17. Change COMP_SUBTYPE to trem
18. Change NAME “C74: Tremolo Plugin” to NAME “yourCompanyName: Tremolo”
19. Open ‘Tremolo-Info.plist’ under Tremolo > Supporting Files in the Project Navigator column on the left
20. Expand the AudioComponents field and change the ‘manufacturer’ to your company name, the ‘name’ to ‘Tremolo’ and the ‘subtype’ to ‘trem’.
21. Almost there!
22. Open Tremolo.cpp from the Project Navigator
23. Scroll down until you see this block of code (line 87):
    
24. Change the arguments of AUEffectBaseMultichannel to (component, 2, 2):
25. Hit cmd+B to build!
26. Xcode should pop up a long list of semantic errors. Ignore.
27. Go to the ‘Library’ folder in your User directory (Users/YourName/Library). If you are on 10.7 and up, Google how to find it
28. Under the Library folder go to ‘Caches’ and delete “com.apple.audiounits.cache” (this will force your DAW – Logic primarily – to rescan all AU plugs)
29. Empty Trash
30. Go back to the Project Navigator, under the ‘Products’ folder right (or control) click ‘Tremolo.component’ and click on ‘Show in Finder’
31. Copy ‘Tremolo.component’ from Finder
32. Paste it in: Mac HD > Library > Audio > Plug-Ins > Components
33. Open up your AU host (you could also use ‘AU Lab’ provided with Audio Tools for Xcode)
34. Done!
35. If you have any problems, recheck the steps (and the pre-requisites)

Hopefully this post is enough to get you started with exploring gen~ and its capabilities so that when the code export feature is out of beta you will be all ready to make the most use out of it!

I have made a few more changes to the code so that the plugin opens with some default parameters and better labels.

And a video of it in action:

Click here to view the embedded video.

You can download the patch, Xcode project and AU plugin from:

I just spent a few hours on this so there is no guarantee that the plugin or the Xcode project will work well on your computer. Moreover, the AU plugin is not the most efficient bit of DSP in the world. Converting a phasor to a sine wave in real time mathematically is not the most CPU efficient method (a wavetable in most cases can work better). Also, from what I understand the AU specification requires all parameters to be within the range of 0 to 1, which I have clearly broken with the Frequency and Phase parameter.

Happy coding and patching.

Do post any questions, comments or criticism.

For those of you who just want to play around with the finished project, there is a download link at the bottom of the article. Despite all of this, I’m still relatively new at Pure Data and the Max language. To those who chime in with corrections or clarifications in the comments, you are most appreciated! If you’re new to PD, make sure you check the comments section for clarifying info provided by generous souls.

Last time, we implemented a three stage filter section with independent LFOs to sweep the center/cut-off frequency of each one. Today, we’ll finish up this patch by adding two last features…an anti-aliasing filter and the ability to record to the hard drive directly out of the patch. If you’re somehow just finding this series of tutorials, or you haven’t finished the previous steps, might I suggest you look those up here? This will also be the last time I’ll remind you to setup your MIDI controller in Pure Data before opening your patch.

First off, disconnect your [dac~] atom from [catch~ output], and move it a bit off to the side for now. Create a new sub-patch – [pd antialias]. Now, because I’m not nearly knowledgeable enough to properly construct an anti-aliasing filter, we’re going to cheat on this part…BIG TIME! Go to the “Help” menu in the program menu bar and select “Browser”.Click on “Pure Data/”, then on “3.audio.examples/” and scroll down to “J07.oversampling.pd”. Double click that patch to open it, then open the [pd 16x] sub-patch. This is an anti-aliasing filter that someone else has built, and we’re just going to use it in our patch. Copy everything in in that sub-patch and paste it into your [pd antialias] sub-patch. Now that we’ve copied this code into our patch, you can close the “J.07…” patch. [There are a lot of other useful patches and code examples throughout the help browser. Explore it sometime to see what cool things you can find.]This filter is currently configured for 44.1 kHz, with a cutoff frequency of 15 kHz. If you’d like to change it to fit 48 kHz, with a higher cutoff frequency first click on the [15000]message atom, then note the values in each of the number atoms beneath [pd buttercoef3]. Identify where each of those numbers have been placed in the string of atoms on the left, because you’ll need to replace them in the proper locations. Change the [15000] message atom to hold the cutoff frequency you would prefer (i.e. 18000), and replace the “22050″ in [/ 22050] with “24000″. Switch out of “edit mode” and click the message atom. The number atoms will update with the values that you will need to use in the atoms on the left (all between [inlet~] and [*~ 1]). Keep in mind that you will also need to change the sample rate in the “Audio Settings” of PD (Pd-extended > “Preferences” on Mac, “Media” menu on PC). Honestly, I’ve just left my filter at the default settings. Close the sub-patch and connect it to the outlet of [catch~ output] and the inlets of your [dac~].

It’s time to start building the record function. Create the atom [writesf~ 1] and connect it too to the outlet of [pd antialias]. Leave lots of space between the two atoms, though. We’re going to need it.

[writesf~] is a command to create a sound file. The argument you provide determines how many channels the audio file should have. We’re not working with a  multichannel signal, so “1″ will serve us just fine. Take a moment to view the “help” file of [writesf~ 1] by right-clicking on it. You’ll notice that this atom requires multiple messages and pieces of data to work properly. It automatically records to the same directory as the patch. Note that you can record to a sub-directory if you like, but you’ll need to create that directory before actually attempting to create a file.

We’re going to set this up to generate a file with a common name, but an iterated number. That way we won’t have to edit the file name in between each recording of a session. Also, we can use [start] and [stop] message atoms alone…but adding a button [bang] for each will make things a little more clear cut for the user, and will help us control some other useful features as well.

Create a [bang] (button style) and connect it to a new [t b b] atom. We’re going to use this [bang] -> [t b b] combo to control the order of operations, so that we don’t need to remember to create our file name and “open” the file before starting the recording. First, let’s build that iterative number with the old stand-by “counting machine” we’ve used in the past; [f 0] <-> [+ 1]. To give ourselves a way to manually reset the counter though, let’s create a [0] message atom and connect it to the cold inlet of [f 0] as well. Connect the inlet of [f 0] to the right outlet of [t b b], then create  a new [start] message atom and connect it to the left outlet of [t b b]. Connect the [start] message atom to the inlet of [writesf~ 1].

So, our counter fires first, then the “start” message is sent to [writesf~]. Remember though, that PD will complete all code in the right-most path first. If we build our file name creation and “open” argument directly out of [f 0], that means that all of that will be completed before the [start] atoms receives its “bang.” Create the atom [makefilename sound%d.wav]. [makefilname] does exactly what it sounds like; it simply generates a file name as a “symbol”. We haven’t spoken about “symbols” at all, but you can think of them as strings of text. They are not recognized as commands, floats or arguments on their own…but you can send a “symbol” to a variable where one of those is expected. If you have a symbol atom (cmd+4) [13] and fed it to a [float], [float] would output the integer value “13″. So, a symbol is not a value, command or argument…it is simply a string of characters…but it can be read as a value, command or argument when applied in the right situations. Let’s continue by looking at the argument we created in this atom, “sound%d.wav”. The “%d” is a variable specific to [makefilename] that will allow you to insert an integer number into the file name string. [For more on the variables for this particular atom, view its "help" file.] If we connect [makefilename sound%d.wav] to our [f 0] <-> [+ 1] counting pair, we’ll end up with a new file name every time we click the [bang] to start recording: “sound0.wav”, “sound1.wav”, “sound2.wav”, etc. This greatly simplifies the record process within a session. You can easily replace the “sound” portion of the argument with your own custom name. If you’ve created a recording sub-directory (we’ll use the directory name “recordings” as an example) within the directory that houses your patch, you would change “sound%d.wav” to “/recordings/sound%d.wav”. That simple.

We’ll feed this filename (with possible sub-directory if that’s how you’ve configured it) to the “open” command required by [writesf~]. Create the message atom [open -bytes 3 $1]. If you read through the help file for [writesf~] carefully, you may already have a clue to what all of that means: “open” tells [writesf~] to create a new file, “-bytes 3″ tells it to create a 24 bit file, and “$1″ accepts the symbol from [makefilename] and uses the character string for the file name. The way we have it set up right now, PD will record the file in whatever sample rate it is currently set to. If you’d like to ensure that you’re always recording in a specific rate (we’ll use 48 kHz as an example), you’ll want your message atom to look like this: [open -bytes 3 -rate 48000 $1]. [If you do this, don't forget that you'll want to adjust the values in your [pd antialias] sub-patch.] Connect this new atom to the outlet of [makefilename sound%d.wav] and the inlet of [writesf~ 1].Next, we need to be able to stop our recording. Create another [bang] (button style), then a [stop] message atom. Connect the [bang] to [stop], then connect [stop] to [writesf~ 1]. Our recording controls are now finished!Let’s finish up by doing a few things to help the user out. First, we’ll make a “recording indicator”. Create a [toggle], connect it to a new [metro 500], then connect that to a [bang] (button style). Let’s make that new [bang] really easy to see though. Open its properties, and change its size from “15″ to “50″. That’ll be attention grabbing, won’t it? Create two command message atoms with sends over “switch”, and values of “1″ and “0″. I want you to connect the first, [; switch 1], to [start]. The other, [; switch 0], should be connected to [stop].Open the [toggle]‘s properties, and set its receive-symbol to “switch”. Test out the recording indicator by switching out of “edit mode”, and clicking on [; switch 1]. The large [bang] should start flashing. Click on [; switch 0], and it should stop.

The only other thing that I would suggest you do now, is go through and add some color to all of the controls that the user should be concerned with. If you open the properties window of any visual atom (bang, slider, radio, etc.), you’ll see the color selections at the bottom of the window. If you want a reference as to what the important controls are, just have a look at my finished project below. To get those nice fancy rectangles behind the text “comments” use the “Put” menu (or shift+cmd+c) to place a canvas in the patch. You’ll need to adjust its size and color in its properties menu. You may run into the issue that it passes over your comment. If that happens, “cut” the comment (the usual cmd+x) from the patch, and paste it in again. you should have no trouble placing the text over the canvas now.

What we’ll be building in this series

We are now done! I hope this project has given you enough confidence in PD to start playing around on your own. If nothing else, you’ve got a fun little noise-maker that might turn into a useful tool in the future. Thanks for reading this far!

If you’d like to compare your finished patch to mine for troubleshooting purposes, or don’t have the time to build this whole digital contraption and just want to play around, you can download the three files here (make sure you keep them all in the same folder). [disclaimer/don't sue us note: These files are provided in support of this series of tutorials. The download and use of them is done entirely at your own risk.]

A major challenge in video games is to record and manage thousands of dialogue lines. In Mass Effect 3, for example, forty thousand lines were recorded, and in Saints Row: The Third, eighty thousand lines were needed. Having to deal with so much dialogue can be extremely time consuming and expensive.  For this reason, I decided to build a prototype of a fictional language dialogue system as a possible solution that would simplify processes and reduce costs. I chose Max/Msp because as a non-programmer, I needed a tool that would allow me to experiment and develop my own rules and behaviors. The high versatility of Max/Msp would let me build the exact design I had in mind.

Click here to view the embedded video.

BACKGROUND

Invented languages have been widely used for fictional characters of stories, movies, and video games[1]. These range from highly complex and fully functional, like the Klingon in Star Trek, to nonsensical gibberish that only intends to produce an illusion of a real language (e.g. Simlish, spoken by The Sims, which does not express any literal meaning, but reflects the character’s emotions through intonation [5]). Darren Blondin posted an excellent article about constructed languages (link: http://www.dblondin.com/092507.html) where he analyzes another approach for creating them, which is based on sound design. For animal-like creatures in Star Wars, for example, Ben Burtt recorded sounds of groans and vocalizations of bears, lions, walruses, and dogs. Then, he classified them into categories (angry, happy, and inquisitive) and built a word list. Next, he modified and processed the sounds with pitch variations. Then he added overlapping layers, and finally created phrases by concatenating the sounds [3].

Design Concept

The FLDS creates phrases based on how languages work, by concatenating and recombining sampled sounds (phonemes and morphemes) to build words. Based on probabilities, it creates series of words that play one after the other, with pauses in between. However, it only produces an illusion of a language and has no grammar rules. It uses different settings of pitch, pace, pauses, and volume dynamics to express specific moods. Also, several layers of sounds can be included to add common elements and bind sounds together, providing a natural feel to the language.

Creating an impression of a real language

Real languages have a phonological aspect that determines the way in which words are pronounced. A phoneme is the smallest unit of sound, such as /s/ or /b/. Phonemes are key elements in this system because they can characterize a language and produce a sense of unity. For example, in the Polish language, the most repeated phonemes are “ch,” “sh,” and “dg”, and add a particular seal to it [11]. For alien or creature dialects, instead of phonemes you might want to use unconventional noises, such as clicks (like in District 9), which would be equivalent to these unifying sounds.

The next key component of the system is a morpheme (or main sound). Morphemes are the smallest units of language with a meaning (e.g. car, or reach) [10]. Words can comprise one or more morphemes. Air and line, for instance, can be combined to produce airline. An aspect of morphemes is that they may vary in size (e.g albatross, water, cow). This is appropriate for the system because concatenating sounds at a smaller level, like syllables, would result in fragmented and disjointed words.

Silences or pauses also play an important role in language. They can either be grammatical, to separate phrases from each other, or non-grammatical, such as breathings or filled pauses (e.g. hmm, eh, and uh). The occurrence of pauses depends on different semantic, phonetic, physiological and or cognitive factors. Breaths, for example, are usually physiological longer pauses. Non-breaths might vary depending on speaking rate [16].

The FLDS simulates a structure based on sounds that would correspond to morphemes as the building blocks of a language, specific phonemes or noise clicks as the unifying elements, and pauses to provide articulation to phrases.

Emotional Speech

“A language, or more accurately, the sensation of a language, has to satisfy the audience’s most critical faculties. We are all experts at identifying the nuances of intonation. Whether we understand a given language or not, we certainly process the sound fully, and attribute meaning—perhaps inaccurate—to the emotional and informational content of the speech”.

Ben Burtt [3]

Apparently, humans can perceive emotional tone in the psychophysical qualities of another person’s voice, and different emotions can be expressed in dialogue depending on combinations of volume, tempo, and pitch contour [8]. The mood settings of the FLDS were based on Juslin’s charts that associate several voice tone behaviors with emotions. The system focuses on sad, angry, and happy moods. It also includes neutral as an emotionless state.

Chart by Juslin and Laukka [9] that summarises main patterns of speech depending on the mood.

Click here to view the embedded video.

HOW IT WORKS

 Overview

The system concatenates events in the following order:

 Morpheme (or main sound) – Phoneme (or unifying sound) – Breath (or filled pause)  – Delay

 Morphemes are compulsory and the other elements are optional. Prior to generating a phrase, the user can configure the settings of these variables and set the mood, which and what percentage of phonemes to include within the words, the percentage of breaths, and the probabilities of obtaining pauses, among others. Then, a sentence length must be specified to define how many times the sequence should be repeated. The mechanism will be explained in more detail below.

Sections

1) Load Folders

Be sure to click on an image to see a larger version of it for any details.

In the first section, the user can load folders with samples classified into main sounds (per moods), question endings, unifying sounds, breaths (or filled pauses), and layers. There is also an option that will automatically load folders of two preset examples.

Under the hood:

The [polybuffer~] object allows to store and reference multiple sound files from a folder. There is one [polybuffer~] per folder.

2) Configure Phrase Settings

The next section allows to configure the phrase settings. The user can choose the mood, percentage of unifying sounds, probabilities for each type of unifying sound to appear, percentage of breaths in a phrase, probabilities of obtaining pauses, the duration of those pauses, and a question ending option.

Regarding moods, there is a slider that goes from furious to excited. If the happy mood is selected, the sequence will include random files from the happy folder. If the user selects a higher degree of happiness, towards excited, the sequence will include less files from the happy folder and more files from the excited folder. Selecting a mood will also automatically set presets for volume dynamics, pitch, and pace.

Note that all samples are randomly selected from each folder.  However, to avoid phrases sounding too repetitive, the system is configured to not generate two equal consecutive morphemes. Each time an [urn] object receives a bang, it evaluates the total number of sound files in a [polybuffer~] and outputs a random number that will never be the same as the immediate previous one, allowing to generate series of numbers without duplicates. A future improvement would be to make this setting optional, in case an alien language is intended to sound repetitive.

3) Create a Series

In order to create a series of MPhBD, a sentence length must be specified that will determine how many MPhBD sequences will be created.  For example, if the sentence length is 10, the system will produce a series of 10 MPhBD sequences, (each one with different values but based on the same settings), obtaining, for example, something like this:

MMphMB-MMph-Delay- MphMB-MMphMph

Under the hood:

Once all variables and events are configured, and the sentence length has been specified, the set button must be pressed. This will create the lists of values correspondent to a series. As each MPhBD is based on random and probabilistic choices, all lists and messages of a series are stored into [coll] objects so that they can be recalled in the exact same way any time the user decides to replay a phrase. In more technical words, set sends the sentence length to a [uzi] object that will output the specified number of bang messages to all of the elements of the series stored in [coll] objects.

4) Play the Series (Unpack the lists, and set the grooves)

When start is pressed, the sentence length is sent to a counter. Each time the counter increases, the correspondent MPhBD values stored in [coll] objects are unpacked. The unpacked values are used to set and play five groove objects: one for morphemes, one for phonemes, one for breathings, one for question endings, and one for layers.

Concatenating sounds and pauses

When the counter starts, the first morpheme is played through the first groove object. Once it is done playing, the groove object outputs a bang (edge bang) that will indicate the next groove to play the next sound. The system is constantly evaluating conditionals to decide which groove should play next.

5. Other

 The other section allows to configure independent and global volumes. It also has additional effects; a Delay and Reverb for the main words, a pitch shifter that uses a variable line, and a Global Reverb to bind sounds together.

6. Layers

This section contains global controls for adjusting layers. Any audio file can be used as a layer underneath the main series of morphemes, phonemes, and breathings. The files are played whenever the [peakamp~] object detects a signal from the main sequence.

Depending on the mood, layers have different settings of pitch and pace (which can be scaled). The angry moods, for example, are higher in tone, shorter in duration, and have little pitch variability, giving the impression of sharp and repetitive attacks. The sad and neutral layers are longer, lower in pitch, and more constant (low pitch variability). The happy moods have a wider pitch range, which means that sounds will have higher pitch variability. Also, in the happy moods, pitch changes are triggered at more irregular times (providing more microstructural irregularity) than in the other moods.

Layer For Robotic Creatures – Sinewave + Granulator

One of the layers is a sine wave that is triggered when [peackamp~] detects signal from the main series of words. The sine wave is altered with a granular synthesis object called [munger~]. Each mood has different settings for the granulator and the decay time of the signal.  Similarly to the previous layer, the angry moods sounds sharper, with fast attacks. Sad has longer grain size, and thus, softer and more legato sounds. Neutral mood does not use a granulator, but just the sine wave at a low and constant tone register. The happy moods have more grain pitch variability, grain size variability, and separation.

Layer with longer overlapping sounds

 The previous layers are synchronized with the main MPhBD words. A different layer of longer sounds can be added that overlaps with the main words. This can be useful to add commonality and bind sounds together.

7. More Settings

In this section, some of the default settings can be modified. The pace and pitch range for morphemes, phonemes, and breaths can be scaled, and pauses can be configured.

Testing

Fifty two participants were asked to answer questions regarding the mood expressed on eight phrases for two different creatures. The first four phrases correspond to a large animal-creature made of pig vocalizations and other wild animal sounds. The other four phrases correspond to a small alien, and were created with my own voice. Participants were asked to match a mood with a sound file.

The results demonstrated some confusion among participants. In the first creature, for example, sad and neutral moods were often confused. However, sad and neutral have similar pitch and dynamic qualities. Both have a low pitch range, low volume, and low pitch and dynamics variability. In other words, both moods share the same range of arousal, which means, these results are not negative. Similarly, in the second creature, angry was confused with happy, which is apparently opposite, but in truth shares the same arousal dimension. This means that one of the major problems that caused confusion was the intrinsic nature of the sounds. For example, in the first creature, the sad samples were made of gorilla breaths. Even though they are softer than the other samples and have elements that are supposed to be associated with sadness (more legato, lower pitch register and volume, more pauses), listening to a wild animal out of context can bring multiple interpretations. In fact, wild animals are usually associated with threat, and that is why they are sometimes used in film to characterize danger [15].

It would be necessary to analyze other aspects of sound that people associate with mood such as timbre. The sound produced by a kitten, or a bird, might be softer than the gorilla breaths. However, note that choosing the sounds based on mood and timbre associations would imply a restriction for creating the right language, as it would not always fit with the visuals of the character. If the dialogue is generated for Godzilla, it would not be appropriate to use bird sounds. In other words, sounds need to be based on the character of the game. Is it possible to “humanize” sounds, regardless of their nature, and fit them into moods? Are pitch, dynamics, pace, and pauses enough?  Probably yes, but in order to prove it, it would be more reliable to test the sound against an image within a context.

Final Thoughts

It might seem ambitious to condense presets of pace, volume and pitch parameters to work generically for various creatures. Each monster will always have its own needs and variants. However, when using Max/Msp, there are endless options that can be added to expand the system and make it more flexible. For example, alternatives for raising or lowering intonation could be implemented (using formant pitch shifting and granular synthesis). The system could also have specific rules for pauses; instead of occurring at random places, they would occur at n% of each phrase (specified by the user). The FLDS would also be more efficient if it had more types of sounds classified into subfolders (e.g. initializing sounds, connector sounds, and ending sounds) and this way, it would provide more smoothness to sentences.

The FLDS is still in a testing and developmental stage. However, although it still has wide room for improvement, it serves to illustrate the potential of Max/Msp as an ideal environment to build your own sound design tools and prototypes.

REFERENCES

[1] Blondin, D. (2007) Language Design for Sound Designers. [Online]. Available from:

http://www.dblondin.com/092507.html

[2] Bridgett, Rob (2009). A Holistic Approach to Game Dialogue Production [Online]. Available from: http://www.gamasutra.com/view/feature/4178/a_holistic_approach_to_game_.php

[3] Burtt, B. (2001) Star Wars: Galactic Phrase Book and Travel Guide. 1^st ed. New York: The Random House Publishing Group.

[4] Clark, Herbert H. & Wasow, Thomas.  (1998) Repeating Words in Spontaneous Speech. [Online]. Available from: http://www-psych.stanford.edu/~herb/1990s/Clark.Wasow.98.pdf

[5] Ellison, B. (2008) Defining Dialogue Systems. [Online]. Available from:

http://www.gamasutra.com/view/feature/3719/defining_dialogue_systems.php?print=1

[6] Gross, Ariel (2012) 80,000 Lines: Three Lessons Learned. [Online]. Available from:

http://www.gdcvault.com/play/1015913/80-000-Lines-Three-Lessons

[7] Isaza, M. (2009) Exclusive Interview with Dave Whitehead, Sound Designer of District 9. [Online]. Available from:

http://designingsound.org/2009/11/exclusive-interview-with-dave-whitehead-sound-designer-of-district-9/

[8] Juslin, P. Västfjäll, D. (2008) Emotional responses to music: The need to consider underlying mechanisms. [Online]. Available from: http://nemcog.smusic.nyu.edu/docs/JuslinBBSTargetArticle.pdf

[9] Juslin., Laukka. (2004) Expression, Perception, and Induction of Musical Emotions: A Review and a Questionnaire Study of Everyday Listening. [Online]. Available from:

[10] McMahon, A. &  Bermúdez-Otero, R.  (2006) English phonology and morphology. In:

Aarts, Bas, & McMahon, April (2006). The handbook of English linguistics. Oxford: Blackwell. 382-410. [Online]. Available from:

http://www.bermudez-otero.com/bermudez-otero%26mcmahon.pdf

[11] Nevin, Kelly (2009) Guide to Polish Pronunciation. [Online]. Available from:  http://www.nevinkellygallery.com/essays/essay-guide.htm

[12] Pereira, Cécile. (1996) Proceedings of the sixth Australian International Conference on Speech Science and Technology. Angry, Happy, Sad, or Plain Neutral? The identification of vocal affect by Hearing-Aid Users. [Online]. Available from:

http://www.assta.org/sst/SST-96/cache/SST-96-Chapter11-p9.pdf

[13] Purslow, Matt. (2011). Mass Effect 3 contains twice as much dialogue as the first game. [Online]. Available from:

http://www.pcgamer.com/2011/10/13/mass-effect-3-contains-twice-as-much-dialogue-as-the-first-game/

[14] Searle, John R.  (2006) What is Language: Some Preliminary Remarks. [Online]. Available from:  http://socrates.berkeley.edu/~jsearle/whatislanguage.pdf

[15] Whittington, W. (2007) Sound Design & Science Fiction. Austin, TX: University of Texas Press. p.96, 97

[16] Zvonik, E. 2004. Pausing and the temporal organization of phrases. An experimental study of read speech. [Online]. Available from:  ftp://193.1.133.101/pub/fred/docs/zvonikThesis.pdf

Special thanks to Naila Burney for this article. You can finder her on Twitter: @NailaAshi and at gameaudiostuff.com and be.net/nailaburney