Recently, Gary Mairs, a filmmaker and faculty member at CalArts, asked me to create a score for a screening of Teinosuke Kinugasa’s landmark silent film, Kurutta Ippēji (aka, Page of Madness” aka “A Page out of order”) in his Film History Class.
Accompanying silent films was a gig I had for several years at the school and enjoyed it so much that I started doing live scoring in other venues as well. Gary’s class gave me the opportunity to create scores (with artists such as Carmina Escobar and the Rough Hewn Trio’s Craig Bunch and Chris Lavender) for many films including The Cabinet of Dr. Caligari, Broken Blossoms, The Smiling Madame Beaudet, Faust and Phantom of The Opera.
The experience with Carmina Escobar was so successful that when the opportunity arose to create a live score as part of the Cha’ak’ab Paaxil Festival in Mérida, Mexico ( and present a workshop on “Structured Improvisation in Film Accompaniment” at the Edificio de Artes Visuales – Escuela Superior de Artes de Yucatán) I took it immediately.
Of all the films I’ve accompanied though, Page of Madness holds a special place in my heart. The initial score that Carmina and I improvised for Gary’s class was a moment that resonated strongly with everyone in the room and I knew immediately that it was something that I wanted to add to my live repertoire. When I moved to NY, I knew that I wouldn’t be able to accompany Gary’s class live anymore, but having the opportunity to do this was a way for me to create some kind of document of that time. So I gladly took it on.
The process I use to improvise for film (and how I teach other people to do it) is too long to go into here, so I’ll save that for a future post. For now, I’ll discuss some technical and non-technical aspects of how I approached the film.
Originally released in 1926, Page of Madness was assumed lost for decades until the director found in a rice tin in 1971. He added a score (that I liked a great deal by the way) and cut about 1/3 of the original film and any known prints are based around the 1971 version. In terms of the cuts effect on the story, I don’t know if the original cut’s pacing was more linear, but I suspect it wasn’t.
As the acting in early silent films was rooted in theatrical acting (with actors making large gesticulations and exaggerated characterizations to play to the people in the back of the room), many silent films haven’t aged particularly well. In that regard, one remarkable thing to me about this film is how well it has held up over time. The entire action of the film takes place in and around a mental institution and the energy all the actors put into the characterizations works exceptionally well.
Avante-garde doesn’t even begin to describe this film which (like Murnau’s The last Laugh) uses no title cards and takes place solely in and on the grounds of an institution. In addition to some jarring visual superimpositions, Page also features a story line that uses a series of flash backs and non-linear narratives that complicate the story. The end result is a film that is nostalgic on one hand and surprisingly contemporary on the other.
For those of you who are interested, I’ve uploaded about 30 mins of the Quicktime film and the score to discuss what, why and how I did what I did.
I used Logic for this project. Here’s a screenshot of the session.
You might notice a distinct lack of midi in the score. The majority of work that I did was created with audio and samples rather than midi, but I’ll explain differences in the video breakdown.
Here’s the film:
Right off the bat, let me say that I approached this completely differently than I would have approached scoring a traditional film.
If I got a gig creating cues for a video or a theatrical release, I would tailor the sounds and experience to audio locked into a film print. As this score would be used to accompany a silent film in a film history class, my goal was to create a score that would simulate a live experience of someone accompanying the film, but I also wanted to add foley and FX to create a depth of experience outside of merely adding music.
Working with Limitations
My original scoring idea when I said yes to Gary was to accompany the film with a multitracked session of my recorded voice using extended techniques like overtone singing. I think it would have been really effective but the lack of an isolated recording environment to record that cleanly nixed that idea ultimately.
After I had put a series of sound effects in, I remembered that I had a source recording of Carmina and I accompanying the same film at another venue and I realized that this might be the best way to tie in the loose live performance feel I was going for with the more orchestrated foley I employed. The source audio of the live recording was fairly clean overall except for some places where the bass frequencies were distorting. If you look at the duodenum track, you’ll see some edits I made to either create space or deal with the frequency issue (I basically moved the offending audio to another track and processed it into something useable).
I used a Quicktime film in Logic to score to and exported the audio. I sent Gary AIF, mp3 files and the Quicktime film so he could see how I imagined the synchronization, but I needed to leave the score open enough that even if the synch fell off that the score would still make sense. Again, for a theatrical release I would have used Midi and SMPTE to have everything synch perfectly but given that this is going to be an audio file that plays through the sound system while the film is screening I decided to try to synch up a few things and leave the rest of it open.
Before I added any music. I spent a LOT of time on foley. I pulled a number of samples from Free Sound dot org, and then spent countless hours cutting things up in Fission (A great 2-track editor from the makers of Audio Hijack Pro) and then further mutating them with endless plug ins in Logic.
Okay, here are some notes from the session from the video.
00:00-00:42 – Synch points and establishing tone
I synched up some audio of a man counting backwards from 1-10 in Japanese with the pre-roll to help Gary synch the audio for the class. I decided to use a temple bell sound to help set the mood and synched that to the page turns.
00:46 -02:09 Silence.
I wasn’t sure if this would actually be a part of the screening. In a traditional accompaniment, you’d typically hear some organist pull an old-time radio drama score over any type of credits or title cards. To me, it’s emotionally not part of the actual film so I left it blank. It also helps people focus on the story.
02:10 -03:05 Title sequence.
The bell motive returns. I added in some Noh Drama type percussion but used a sparse rhythmic motive. In the background some reverse guitar loops begin.
Again, most of the initial work done was on foley and placement. Once I added in the music track then it was a matter of balancing the mix and placement of what I had already done. I took out about 25% of the foley work I’d done to make space for the score.
I decided to treat rain almost as a character in this film. I see it as the truth and realization that the husband doesn’t want to face through the whole film. It’s a psychological foreboding of what’s he’s been avoiding and ultimately acceptance of how things are. I use multiple rain loops in the beginning but a single rain loop I created runs in the background of the whole film and continues past the last scene before it cuts to the last image when the sound fades to silence. It’s a subtle detail but one that adds something to the environment.
04:03-05:06 Dream state and Transition.
While I would need to discuss the actual improvisation process in greater depth than I can here, I should discuss at least one aspect of how Carmina and I approached this. The method of improvisation employed here is structured improvisation. By that I mean that we have cues in the film of things that we’re going to do but they might not be specific. It might be something like, “When we get to the first fight I’m going to add a bunch of distorted guitars and lets create a big loop texture.” or it might be a specific melody or rhythmic device. In other words, we have specific cues of things that we work towards we keep an emotional marker in mind but we vary how we get there and what we’re going to do once there. For this scene we wanted something dreamlike and other worldly, so I went with a repeating figure while Carmina sang long tones over it.
Percussion was added to the loop to act as a foreshadowing of the dancing sequence to follow.
05:07-05:06 Reality and Creating the Asylum
The gate door closing was part of a transition to bring the viewer back to our reality and show that this is how the inmate views the world. There are a number of samples that all run during any of the asylum scenes to help set a claustrophobic tone of the institution. After the introduction, the only time the asylum samples aren’t played is during the outdoor scenes.
Minor percussion and storm effects were added to Carmina’s loops to help build tension.
At 06:37, I stopped the percussion to highlight the fact that what this woman hears is outside the rhythm of the institution and is an internal force that she is compelled to interact with. Carmina’s voices do a brilliant job of conveying that idea of multiple voices fighting for attention.
At 07:27 or so, exhaustion gives way to the sounds around the dancer and reality starts to envelop her again. Carmina sings a variation of the earlier melody as a motif. In the performance, I still have a loop I’m fading out on guitar and the low drum sound is me hitting a road case in the reverberation of the space.
At 08:01 or so – I begin to add in a series of samples of heavily affected backwards speaking to represent the voices the wife hears. I use this sound as the general sound of insanity in the asylum, so it comes to the foreground when the wife is on screen but stays back most of the time as another part of the environment.
08:15 The Husband Enters.
All of the characters have a theme except for the husband. I wanted to have the husband be a character that is adrift in this world around him. I thought it would create a silence to contrast all of the other sounds against.
I tried to remain very aware of the spaces in the film. Even though there aren’t any spaces of complete silence (until the end) I wanted to have sparse moments to contrast the rest of the film against.
We sidestep the asylum briefly to see how the wife got to where she is today. Musically, I try to keep things open except for the mob scenes. I begin to build a loop texture here on guitar. The melodies and counterpoint are based around some Hirajoshi-inspired ideas. I wanted to make sure that the percussion and melodic material I was providing has small hints of music inspired by traditional Japanese music while applying those ideas in a very western way. The end result is something sonically that’s difficult to put your finger on.
19:31 The build up
There are two fight sequences that Carmina and I knew that we wanted to make big sonically. In the scenes leading up to this, I slowly began to make the loops more active and dense to show the tension underneath the surface of the Doctor’s routine walkthrough. At 19:31 the textures start evolving in a different direction and slowly moving to the disturbance at 24:59 that leads to the full on freak out at 27:30 or so. After that we bring it down and create another plateau to build from again later in the film.
The process was really pretty simple. Create an environment and then remove absolutely everything that didn’t have to be there. I should also mention that had Carmina and I tried to play really tight specific cues to the original film that flying in a full performance never would have worked. The conversation that the two of us had during the film was something that would have been impossible to replicate completely, and aesthetically I really liked the idea of the music just being a part of a stream that flows along with the film.
Have a plan B
One last thing I should mention is that the process of actually creating the stereo file and synched video was no picnic.
For some reason, the Audio bounce in logic failed every time I tried it. It would run for 7-8 hours and I’d have to force quit the file.
What I decided to do instead was use the aforementioned Audio Hijack Pro, to record the playback output of the logic file. Once I had a stereo file. I edited it in fusion to start at the 3 count lead in and imported the audio and video in iMovie and synched the hits there and output the Quicktime film. I would have been REALLY stuck without Audio Hijack Pro, so I’m grateful that (3 days and 4 real time bounce attempts later) that I came up with that workaround.
All in, there’s about 40-50 hours of work in it.
There’s a few different things going on sonically in the second half, but this post explains a lot of the reasons behind what I did what I did, and perhaps that’s helpful, insightful or just interesting to some of you.
If people are interested, I’ll post the second half at some point and offer some more observations on what was done and why.
Until next time. From the earlobe of Hurricane Sandy – stay dry (and thanks for reading)!
p.s. For the tech oriented amongst you. Guitar sounds were FnH UltraSonic–> Digitech Space Station–>Apogee Duet–>macbook Pro->AuLab–>Pod farm–> Sooperlooper (Both controlled by a Line 6 Mark II pedal board) –>Atomic Reactor amp.