When I first attended the class, to be honest, I was quite lost.  Musically, my background probably extends back to the days singing along with the Saturday morning cartoons.  I did not even know the concept of music expression existed in such richness and deepness.  That is part of the reason that drove the curiosity for this experiment.  I wanted to see firsthand how these expression concepts extended itself to a casual listener like me.  I was particularly drawn to the KTH system as it seem very well developed and most friendly in terms of allowing us to explore and play with the concepts.  Initially, I wanted to use it to create my own expression from a predetermined piece, but soon realized that was easier said than done, and could possibly create erroneous results should I not utilize the program properly.  That was when I did some digging and found some of Bresin’s work where he actually used KTH rules to create emotion based musical expression.  On a sort of side note, it is interesting to look at preferences and emotions elicited from these adjustments because of my speaker design background.  Dr. Toole of Harman research had a groundbreaking paper in the mid 80s that suggested preferences in speakers were not personal based like food or wine… but rather it was based on the speaker that produced the flattest response.  As in what you put in is what you get it.  So its interesting to see if such a concept can be extended to music expression. 

This is a snapshot of a Bresin presentation he made during the summer of 2005 where he points to the various characteristics of each emotion.  For example, we see– (read some examples)

In a bit of a zoomed in example… here is SADNESS. 

In his experiment with expert listeners (music students of the local university), and found high correlation between a lot of the emotions.  Particularly Anger and Happiness. 

So before I proceed– I want to make this presentation a little interactive.  (Explain a little regarding set up)

None / Minimal (for fun) / Basic (1-2 years) / Intermediate (3-4 years) / Advanced (5-6 years) / Expert (7+ years)

Some did sense tension.. Some mentioned the abrupt tone attacks as stated in the earlier chart.. Expressed it as rather than “anger” “I think it's kind of creepy, it sounds kind of lonely but kind of like a music box, the sound certainly wasnt happy. it was more like a timid, melancholy frustration.” “Minor key, but because notes are in higher register and the tempo is not as slow, it evokes tenderness instead of sadness.” “This piece conveys fear to me because of the minor chords used and the steady, quick tempo throughout the piece. The tempo of the song reminds me of the anxiety one feels when afraid.”

Do sense the slower tempo compared to earlier piece.. Conveying more “anxiety”.. Did notice tempo variability and lower sound level, but felt the tempo was slower.. This piece is slower than the first one, which makes the song seem sadder, like someone is moping or mourning or trying to pick himself off the ground but is having a hard time doing so. At first, you think the slower tempo may make it feel kind of sad, but I got more of a feeling of boredom than sadness. Notes and slower tempo make it more deliberated. It puts more emphasis on each note, which buys time to make each one noticeably minor-key sadness

About a fourth saw this as no expression.. A lot of the ppls who added comments regarding their “happy responses” attributed this in comparison to higher tempo and louder dynamics than the previous two pieces.  Even the people who noted “fear” noticed these characteristics– felt like it reminded them of being in a scary movie.

This is probably the most successful trial.  Some called this “Anger” because the “quicker temp and frilly playing” reminded of listeners of “people bickering”.. Again, “briskness implies the most emotion.. And anger seems to be the most emotional of these” “Some feel that the faster pace, erratic nature of the piece suggests a frantic catatonic state.  Evokes fear.”

Person with minimal experience described it as “light, gentle, and smooth– very tender!” Described as similar to #2.. Similar emotions.. Understandable if we flip back to compare tenderness and sadness on the original chart..

“This piece conveys anger to me because of the accented staccato notes that sound like the person who's playing the piece is hitting the keys really hard.” “the dynamics sound a bit different in this one, a little louder and more direct emphasis, which makes me feel like someone is arguing, or trying to make a point” “They way the keys are pressed by the finger there is a push then a release, indicating anger. Steady tempo” “Full-out rage would be sharper, faster, and a touch stronger. This is more of a seething anger.”

These next few slides compare the beginner – intermediate data versus advanced and expert performers.  We do not see a huge difference in results and can conclude that the advanced and expert listeners do not hold a special advantage.  It was noticed however, that the advanced and expert players were able to articulate their thoughts better and  have a larger venacular. 

Again, more the pattern is shown.

Just to round out the rest of the data. 

Correlating the direct emotion had low results, but upon closer review of the comments given by listeners, these comments did correlate correctly with characteristics of certain emotions (ie. Fast mean tempo for happiness and anger).  We have to wonder about the directness of the questions and the affect of relativity as each listener listened to tracks all at once and what they heard before might influence their opinions.  For example, some listeners noted the “no expression” track as “happy” just on the basis of the faster tempo compared to the two tracks before it. Experience was seen to play little role in the experiment as seen by the previous few charts.  Anger needs to be looked at as it was mistaken as sad and fear a lot of the time.  While some of the smaller characteristics of “Anger” did show up in listener comments, they were not the main characteristics of that emotion.  Further evaluation must be done, in either the methodology of the experiment or in the expressive parameters adjusted for “Anger.”

Binary approach to get elicit better response… ie. Does this sound more sad or fearful? Rather than give an MC choice Utilization of different pieces to get a braoder range or results as piece’s natural feel may lead to a lean in results Given reference piece to compare may give people a solid base.  Analogy to comparing cars.  But then the question arises, that when you go see a performance, there is not base to go off of– you just respond to what you hear Which leads to the next issue to look at...maybe we need real performances of these pieces to elicit better responses?  Someone noted that the MIDI performance sounded mechanical by nature and so this could affect the way people interpreted these expressions.. Could be the general nature of the KTH rules.. Need more optimization.. Try again when newer variations of the rules..