Evaluating Human-Computer Co-creative Processes in Music: A Case Study on the CHAMELEON Melodic Harmonizer

Abstract

CHAMELEON is a computational melodic harmonization assistant. It can harmonize a given melody according to a number of independent harmonic idioms or blends between idioms based on principles of conceptual blending theory. Thus, the system is capable of offering a wealth of possible solutions and viewpoints for melodic harmonization. This study investigates how human creativity may be influenced by the use of CHAMELEON in a melodic harmonization task. Professional and novice music composers participated in an experiment where they were asked to harmonize two similar melodies under two different conditions: one with and one without computational support. A control group harmonized both melodies without computational assistance. The influence of the system was examined both behaviorally, by comparing metrics of user-experience, and in terms of the properties of the artifacts (i.e., pitch class distribution and number of chord types characterizing each harmonization) that were created between the two experimental conditions. Results suggest that appreciation of the system was expertise-dependent (i.e., novices appreciated the computational support more than professionals). At the same time, users seemed to adopt more explorative strategies as a result of interaction with CHAMELEON based on the fact that the harmonizations created this way were more complex, diverse, and unexpected in comparison to the ones of the control group.

Keywords: conceptual blending; creativity evaluation; creativity support tools; melodic harmonization; musical harmony.

Figure 1

The two melodies used in the harmonization task. The upper melody (“Menexedes kai Zoumboulia”) was employed in the simple harmonization task whereas the lower one (“Lullaby from Southern Italy”) was used for the computationally-supported harmonization. Arrows indicate the requested harmonic rhythm.

Figure 2

The online CHAMELEON user interface. The melody for harmonization together with the requested harmonic rhythm (i.e., when chord changes should occur) and important harmonic notes (i.e., notes that CHAMELEON will prioritize over others when selecting suitable chords for the underlying melody) were shown at the top. The options for harmonization were provided below. The user could select from the available harmonic styles using a drop-down menu and use the radio button at the bottom to opt for either a single-idiom harmonization or a blend between two different idioms. Some harmonic styles include more than one harmonic mode that the user could chose from. In addition, there was the option to blend two selected styles in different tonalities using the tonal difference drop-down menu. Finally, the interface allowed users to select type of voice leading. If no voice leading was necessary the system output chords in root position. The other option performed a rudimentary voice leading in the bass voice and in the intermediate voices.

Figure 3

Boxplots of the Overall Creativity Index (OCI) in the two experimental conditions for both participant groups. Mean OCI simple for composers = 5.8, Mean OCI simple for students = 5.57, mean OCI comp. supported for composers = 4.7, mean OCI comp. supported for students = 5.6.

Figure 4

Boxplots of the seven common questions between the two experimental conditions for composers and students.

Figure 5

Boxplots of the responses on each task between groups of participants.

Figure 6

Boxplots of the three harmonic features for both the simple and the computationally supported task.

Figure 7

Boxplots of the differences between harmonizations calculated based on the three features for the two conditions of the experiment. Median difference in No. of GCTs = −0.014, No. of chord types = 0, and Pitch Class Profiles = −0.029.

Figure 8

Boxplots of the differences between harmonizations and the typical solution calculated based on the three features for the two conditions of the experiment. Median difference in No. of GCTs = −0.028, No. of chord types = −0.056, and Pitch Class Profiles = −0.013.

Figure 9

Two-dimensional spatial configurations from Multidimensional Scaling on the distances based on (left) the number of GCTs, (middle) the number of GCT types, (right) the pitch class profiles. The red squares represent the implied harmonizations for the two melodies, the white circles represent the 25 simple harmonizations and the black triangles represent the 25 harmonizations that resulted from the computationally supported task.

Figure 10

(A) Left: the beginning of the second variation of the original piece; right: the Whole Tone and Kostka-Pane blend with five semitones tonal difference by CHAMELEON. The harmonic sequence is transposed one perfect fifth up from Am to Em and although there are differences in the spelling of enharmonic notes, in voice leading and some added notes, the harmonic backbone remains the same as the one suggested by CHAMELEON. (B) Top left: the beginning of the original piece; top right: measures 23–25; bottom: the Jazz harmonization by CHAMELEON. It can be seen that the tonic major opening chord (I) also appears in the Jazz harmonization as a I^7♯. The 3rd measure of the Jazz harmonization contains a downward chromatic movement in the three top voices that was adopted by the composer in measures 23–25 of his composition.