Chris Rutkowski, Composer/Pianist

To follow this thread from the beginning, follow this link.

Here are all three movements of the Sax Concerto gathered together in one place.

Concerto for Alto Saxophone and Wind Ensemble
The Indiana University Wind Ensemble, Stephen Pratt, conductor
Thomas Walsh, alto saxophone

1. Rage
2. La Luna d’Oro
3. NOLA

To follow this thread from the beginning, follow this link.

Step 5: Attempt to remove the boominess of the bass drum in the original audio.

The challenge: The audio was mixed down to stereo, and thus has all the instruments on it, not just the drums. Therefore, removing the frequencies where the boom occurs in the drum will also remove those frequencies in other instruments.

As it happens, the boom seems to be most pronounced at around 130 hZ, which is around B an octave below middle C. That’s a very prominent frequency for the bass instruments: trombones, bassoons, tuba, et al. I used two different kinds of equalization (EQ) to mitigate the boom. First, a mild shelving (-7dB) EQ at around 62 Hz. (quite low) to lessen the very low part of the bass drum. Second a more surgical EQ at 138Hz with a parametric EQ around that frequency to remove the most resonant part of the boom. I think the result is a considerable improvement, although when played through a system with a sub-woofer, it’s still a bit too boomy for me. I’m not sure much more can be done, though, given the limitations imposed by the original audio.

Here’s a link to the latest EQ of the third movement.

To follow this thread from the beginning, follow this link.

Step 4: To add a virtual tuba part to the existing pure audio and trombone virtual instrument tracks.

The Challenge: Importing another Standard MIDI File from the score in Finale brings along with it the midi tempo commands from the Finale playback, thus nullifying all the hard work done in the beat mapping.

This was a setback. I decided that it was quicker just to enter the tuba part note-by-note in the sequencer in Logic than to try and import the complete part from Finale. Otherwise, I’d have to spend a lot of time editing the Finale part to make sure no extraneous data gets into the Logic file. I’ve found this to be a common procedure when using tech-doing things the long way is often the most efficient and beneficial. E. g. in this case, by re-entering parts after having written them a year earlier, I relearn them and can more effectively lead the next rehearsal. (Bach did a similar thing when he would copy out scores by Telemann or Vivaldi-by doing so he could learn the music in the most comprehensive manner.) The tuba part is very active and rhythmic, and has to be nailed in order to keep the groove, so this accomplishes many goals at once, and thus the long way is actually the shortest.

Another challenge: I created a beat map correspondence for the first part I imported, the trombone, but the tuba plays in large areas where the tbn doesn’t. So a beat map has to be created for those areas. But because creating a beat map actually creates small areas of different tempos, the bars that follow are altered and have to be redone. I had to redo the beat map for about a week. I think I have a work around though: make sure that there are instrumental parts in the sequencer for every bar that you’re going to add them to the pure audio before you beat map. After that, other instrumental parts can be added on top without having to beat map them, and you only have to do it once.

At this point, I have the low brass virtual instruments added to the audio, and am quite happy with the result. In the pure audio, the drums were a bit too present, and the bass (i.e., the low brass) weren’t properly balanced. Now, with the active bass parts restored, the groove is much stronger, and the piece considerably more exciting.

Here’s a link to a nearly complete comp of the audio enhanced by virtual tuba and trombones. Pay attention to the bottom. A further challenge: the bass drum is too boomy. I’m trying to ameliorate this with EQ, but as the audio was mixed down to stereo, I can’t isolate the drums and EQ them individually. More later…

Audio with a before-and-after comparison coming soon!

To follow this thread from the beginning, follow this link.

Step 3: To add a virtual (synthesized) trombone part to the pure audio file.

So far, I’ve striped the audio file with a bar and beat timeline, that was not present in the original recording. As there is rubato and natural fluctuation of the beat in a live performance, this cannot be done mechanically, but direct correlations must be made on a bar by bar basis. This was accomplished by using the beat mapping function of Logic described in the previous post.

The first challenge was to convert the trombone part from the Finale file into a SMF (Standard MIDI File) which can be read by the sequencer in Logic. In the last post I mentioned that this was more complicated than one would hope. Finale has its own sequencer embedded in the program to allow for its own (quite good!) audio playback of the notation. However, the individual files seem to contain the instrument list of the entire score, and so one must first delete all the extraneous instruments there before converting. After this was done, only one track was created by Logic upon importation. Another problem was discovered after importing. There were cues in the score that were, of course, treated no different from the notes meant to be played by the computer. These also had to be deleted.

When the trombone part was placed into Logic, the process of fine tuning began. Beat mapping had to be considerably refined, with constant reference to the written score. Working with the synthesized trombone required some technical finesse. For example, the timbre of the instrument varies with the dynamic of the attack. This is called velocity in MIDI (values of 1 to 127), and in the trombone patch, the timbre changed from a mellow and warm sound to a brassy one at around velocity 85. However, there is a trick in which the dynamic of an attack can be made independent of the velocity of the attack, and that is use the independent volume (loudness) controls in the DAW. Thus, one chooses the timbre one likes at a particular velocity, and if it’s too quite, just turn up the volume for that note.

I’m quite happy with the first results from the enhanced audio. The third movement of the piece is called “NOLA”, which is the familiar acronym of “New Orleans, Louisiana”. I used authentic New Orleans rhythms throughout, and there is percussion from beginning to end. Originally I used 4 percussionists, but coordinating the parts into a seamless groove was difficult. I then reduced all the parts except the timpani to a single drum set, and this worked like magic, providing a template for all the highly rhythmic instrumental parts to relate to. For the recording, we placed the drummer in the center of the stage, next to the conductor. The problem was that I couldn’t hear enough of some instrumental parts, particularly the bass parts-tuba, trombone, bari sax, bass and contrabass clarinet, bassoons and contrabassoons. As the recording was mixed directly to stereo, I couldn’t do the obvious by simply turning up the volume on the mics that recorded the bass parts. Hence the high tech fix of doubling those parts with synthesized ones in Logic.

I’ll try and post some before and after audio soon. In the meantime, I’ve moved on to adding the tuba part, which lead to a new challenge. That’s for the next post.

To follow this thread from the beginning, follow this link.

Step 2: Create a bar and beat timeline for the audio of each movement of the Concerto.

The Problem: Audio files in a DAW are displayed as waveforms in a timeline, which shows where audio events happen in time. If the audio is recorded without a metronomic click track, then there is no direct correlation between a given downbeat in the audio and the bar/beat timeline used for sequencing MIDI instruments. Thus, in order to add MIDI instruments to the Concerto audio, (to correct dynamic imbalances in the recording) a bar/beat timeline must be created and correlated with events in the audio.

The solution: Logic has a function called “beat mapping”. What it does is adjust particular beat positions in the bar to events in the audio by calculating small changes in the tempo-e. g., the basic performed tempo of the third movement was 105 beats/minute, so if a bar or beat is rushed, Logic will insert a change of tempo for that area when a particular beat is linked to a particular rushed note.

How does it do this? First, I had to recreate the entire metric structure of the score in the Logic file, i.e. changes in time signatures where they occur. Beat mapping displays the audio waveforms under the bar/beat map. The next step is where the magic happens-Logic analyses the waveform for “transients”, which are essentially spikes in volume at the attack points of given notes. Referring to the score, one can draw in a link between a given beat and a particular transient, thus giving Logic two points in time with which to calculate a change in tempo.

Voila! There is now a grid imposed on the audio which can be used by the sequencer to trigger virtual instruments. The first such instrument I used was a trombone to beef up the bass in the 3rd movement-but more about that in the next post.

One last preliminary: I had to get score information on what notes to play into Logic’s sequencer from the score. This was not as easy as I had hoped, but victory was had at last. Finale will convert the score information into a Standard MIDI File of note on/off information, which can be read by the sequencer, which, in turn, tells the synthesizer when to play a given note and how long. Unfortunately, when I made a SMF of the trombone part, it included a lot of other instrument information, and caused Logic to create many more than the one track needed. The solution was provided through a long chat with Finale tech support (not toll-free, but that’s what Skype is for!). Apparently, when the part was extracted from the score, it kept the instrument list (for Finale’s own playback) from the score. I simply had to delete all the extra instruments in the Finale file before saving it as a SMF and importing it into in Logic. The moral: everything has to be massaged….

To follow this thread from the beginning, follow this link.

Step 1: Assembling the best takes of the Concerto recording into the best compilation possible. Basically, this involves cutting the takes up into the best performances of any given passage, then gluing them together into a seamless whole. The length of the segments varies greatly, sometimes only a beat long, if an attack point was better in one take than another.

This is one of the most common applications of hard disc recording programs, and easy to do in Logic. For me, I find Logic easier to work in than Pro Tools. Segmenting and trimming audio regions in Logic is intuitive and comfortable, as is finding good places to merge audio files. Cross fading between the audio regions that are selected is essential, and is very elegantly done in the program. Often I’ve had a cross fade rejected by PT, but not once in Logic. A very passable comp of all three movements was accomplished pretty quickly.

Here’s a link to a decent comp of the first movement, “Rage“.
Here’s one to the second, “La Luna D’oro”.
I’ll leave the audio of the third movement for a future post, as I’ve already taken the first steps in enhancing the recording, and like the result better than the original.

As the recording was mixed down to stereo in the recording process, nothing could be done with balance-at least not easily. This will be the subject of the most ambitious element of the project-adding virtual instruments to the audio recording to address dynamic imbalances. The problem occurs when a click track isn’t used in the original recording, therefore there is no bar ruler in the recording file itself. Using a click track was one of the revelations when I returned to the recording studio last year. Playing jazz with a terrific rhythm section and a click seemed counterintuitive and intrusive, but it is extremely useful when editing the recording, and especially so if you want to add sequenced instruments.

Using automated volume allowed me to adjust dynamics between sections, and went a long way to enhance climaxes and movement toward them. But adding a bar and beat structure to the audio was needed in order to correct imbalances between in the audio by adding synthesized instruments. That will be the subject of the next post.

I’ve been involved in what I’ve called “Artist-Driven Technology” for several years. (Here’s a link to the earliest post, and here’s one to the whole category.) I’ve begun a project that is a perfect example of the technological process merging seamlessly with the artistic one, so I thought I’d document some of the landmarks.

At the core is a piece I composed, my Concerto for Alto Saxophone and Wind Ensemble. Basically I am editing and enhancing a recording of it by the Indiana University Wind Ensemble with soloist Tom Walsh. I’ll be using Logic, Apple’s Digital Audio Workstation (DAW), and it illustrates the new harmony of musicianship and technology that I am really excited about. The project has many phases, each one of which is a career in itself. I’ll begin with some backstory, and subsequent posts will deal with specific tasks and elements as the project unfolds.

The Concerto was commissioned by four universities, and has been premiered by two, reviewed favorably, with two more premiers planned for the coming season. As I get deeper into the tech side, I feel strongly that the fact that I wrote this music for a large acoustic ensemble to be performed on the concert stage informs the tech with a profound layer of meaning that it wouldn’t have otherwise. Hence the new form of life: Artist-Driven Technology.

Technology was involved very early in the process, when the score was copied in Finale. Here’s a link to the first page of the score. Because the third movement involves a strong New Orleans groove, I made a synthesized recording of the movement in Reason to help the ensemble feel the syncopations. It’s possible to make a good recording within Finale, but doing the drums there poses a lot of challenges. Reason contains a very nice drum synthesizer, and I was able to write the parts in it quickly, so that was the better option. I converted the notation in Finale to a Standard MIDI File easily, and simply imported it in to Reason. With a few tweaks I had a nice demo. I emailed an mp3 to the conductor, and he played it for the ensemble, as well as forwarding the email to the players.

In November, 2008, the IU group did a studio recording of the piece, and I got a CD of the raw takes. My first task was then to assemble the best takes into the best whole. That’s the subject of the next post.

My recording of the beautiful Brazilian tune “So Many Stars” with Heather Ramsey was montaged on youtube by MELEGORM.

Dudamel conducting his high school kids in Venezuela playing Shostakovich 10, and in conversation at the TED conference.

I’m reposting (for the third or fourth time….) this link to a profile of the Japanese multi-media artist, Takagi Masakatsu. His work is very beautiful, and as a video artist, composer, performer of the sound track, and the technologist behind every element; he is the perfect model for the creative individual who wants to take charge of the technology and produce first rate work on a commercially-available desktop.