Date post: | 09-Feb-2017 |
Category: |
Education |
Upload: | multimediaeval |
View: | 179 times |
Download: | 7 times |
The C@merata Task at MediaEval 2015: Natural Language
Queries on Classical Music Scores
Richard Sutcliffe, University of Essex Chris Fox, University of Essex
Deane Root, University of Pittsburgh
Eduard Hovy, Carnegie-Mellon University Richard Lewis, Goldsmiths, University of London
2
Outline Applications
Task Design
Passages and Divisions
Query Types and Music Scores
Gold Standard
Evaluation Metrics
XML Formats
2015 Campaign
Results
Conclusions
3
C@merata stands for: Cl@ssical Music Extraction of Relevant Aspects by Text Analysis
4
C@merata Task There is a series of questions with required answers. Provided Question:
• A short noun phrase in English referring to musical features in a score,
• A short classical music score in MusicXML. Required Answer:
• The location(s) in the score of the requested musical feature. We call each such location a passage.
5
Applications Linking musical references to scores: e.g. link ‘the first movement coda’ to the final bars of a work Allowing experts to search scores interactively using phrases e.g. ‘Valhalla Theme in Das Rheingold’ brings up a matching passage Helping students to learn music theory e.g. ‘perfect cadence’ brings up an example in a score
6
Task Design There are 200 queries. We saw earlier that we have: Provided Question:
• A short noun phrase in English referring to musical features in a score,
• A short classical music score in MusicXML. Required Answer:
• The location(s) in the score of the requested musical feature. We call each such location a passage.
7
Bach BWV 1047 Andante with sample questions and answers
8
Q: dotted minim F#4
A: [ 3/4, 1, 65:1-65:3 ]
Q: F4 crotchet in the oboe
A: [ 3/4, 2, 64:3-64:4 ]
Q: minim A2 in 3/4 time
A: [ 3/4, 1, 62:2-62:3 ], [ 3/4, 1, 64:2-64:3 ]
Q: chord D2 E5 G5 in bars 54-58
A: [ 3/4, 2, 57:1-57:1 ]
Q: quavers F3 A3 followed by crotchet A4 in the violin
A: [ 3/4, 1, 57:2-57:3 ]
Q: four quavers in the violin against a minim in the bass clef
A: [ 3/4, 1, 62:2-62:3 ], [ 3/4, 1, 64:2-64:3 ]
9
Passages and Divisions We specify a passage as follows:
• a time signature, • a divisions value • a start bar and beat • an end bar and beat Example:
[4/4,1,1:1-2:4] time signature is 4/4. divisions value is 1 so we count in crotchets (quarter notes) Passage starts in bar 1 before the first crotchet (i.e. 1:1) Passage ends in bar two after the fourth crotchet (i.e. 2:4) Thus passage consists of the two complete bars (measures) numbered one and two.
10
What is a Divisions Value? The number of beats into which we divide a crotchet. A suitable value depends on what we wish to demarcate. Working in whole crotchets: Divisions = 1
Working in quavers: Divisions = 2
Working in quaver triplets: Divisions = 3
Working in semiquavers or quavers or triplet quavers: Divisions = 12 !!
With Divisions = 12:
one crotchet = 12 beats
one quaver = 6 beats
one quaver triplet = 4 beats
one semiquaver = 3 beats
Note: MusicXML uses the divisions concept and we got the idea from there.
11
Query Types In 2014 there were twelve types of query:
Type No Example
simple_pitch 30 G5 simple_length 30 dotted quarter note
pitch_and_length 30 D# crotchet perf_spec 10 D sharp trill stave_spec 20 D4 in the right hand word_spec 5 word "Se" on an A flat
followed_by 30 crotchet followed by semibreve melodic_interval 19 melodic octave
harmonic_interval 11 harmonic major sixth cadence_spec 5 perfect cadence
triad_spec 5 tonic triad texture_spec 5 polyphony
All 200
12
In 2015 there are eight types of query, each modified in various ways:
Type No Example
1_melod 40 D4 minim
eighth note in measure 9
1_melod qualified by
perf, instr, clef, time, key
40
trill on a quaver A G# in the Cello part in measures 29-39
sixteenth note C# in the left hand half note E3 in 2/2
sixteenth note G in G minor in measures 1-5
n_melod 20
F# E G F# A Do Mi Do Sol Do Mi Sol Do in bars 1-20
twenty semiquavers five note melody in bars 1-10
n_melod qualified by
perf, instr, clef, time, key
20
two staccato quarter notes in the Violin 1 crotchet, crotchet rest, crotchet rest, crotchet, crotchet rest, crotchet,
crotchet, crotchet, crotchet, crotchet in the Timpani melodic octave leap in the bass clef in measures 70-80
G4 B4 E5 in 3/4 rising G minor arpeggio
13
1_harm poss qualified by
perf, instr, clef, time, key
20
eighth note chord Bb, C, E chord of D minor in measures 109-110
harmonic minor sixth in the Violas dotted minim chord in the left hand
texture 6
monophonic passage homophony in measures 1-14 polyphony in measures 10-14 Alberti bass in measures 0-4
follow possibly qualified on
either or both sides by perf,
instr, clef, time, key
40
quavers F4 E4 in the oboe followed by quavers E2 G#2 in the bass clef quarter note minor third followed by eighth note unison
C followed by mordent Bb chord C4 G4 C5 E5 then a quaver
three eighth notes in the Violin I followed by twelve sixteenth notes in the Violin II in measures 87-92
synch possibly qualified in
either or both parts by perf,
instr, clef, time, key
14
four eighth notes against a half note crotchet D3 on the word “je” against a minim D2
four staccato quavers in the Violoncello against a minim chord Ab3 C4 F4 in the Harpsichord
All 200
14
2014 Scores
Work Staves Scoring Lang
bach_cello_suite_1_bwv1007_prelude 1 vc Amer. bach_chorale_24835b3 4 SATB Eng.
bach_chorale_507b 4 SATB Amer. bach_minuet_in_g_bwv_anh114 2 hpd Amer.
carissimi_o_felix_anima 3 SAB Eng. charpentier_te_deum_preludium 5 2 vn, va, 2 vc Amer.
corelli_allegro_tr_clementi 2 hpd Eng. cutting_galliard_11 1 lute Eng.
dowland_earl_of_essex_measure 1 A Amer. lassus_psalm_50 3 SAB Eng.
lully_andante 3 2 vn, vc Amer. monteverdi_lasciatemi_morire 5 SSATB Amer. purcell_fie_nay_prithee_zd10 3 SSA Eng.
scarlatti_a_se_florindo 3 S, hpd Amer. scarlatti_k466 2 hpd Eng.
tallis_all_praise_to_thee 1 A Eng. telemann_taenzchen 2 hpd Amer.
telemann_twv33_21_tres_vite 2 hpd Eng. vivaldi_concerto_rv299_largo 3 vc, hpd Eng.
weiss_sonata_34_prelude 2 lute Amer.
15
2015 Scores
Work Staves Scoring Lang
bach_violin_sonata_no1_bwv1001_presto 1 vn Amer. bach_violin_sonata_no2_bwv1003_andante 1 vn Eng.
mozart_piano_sonata_k545_m1 2 pf Amer. mozart_12_horn_duos_k487_no1_allegro 2 2 hn Amer.
bach_cpe_duet_flute_violin_h598_andante 2 fl, vn Amer. clementi_sonatina_op36_no1_v2 2 pf Eng.
scarlatti_k30 2 hpd Eng. bach_marcello_bwv974_adagio 2 hpd Amer.
mozart_an_chloe_k526 3 S, pf Amer. telemann_ceciderunt_in_profundum 3 SSB Amer. purcell_fantasia_no10_4_parts_z741 4 4 viols: tr, 2 t, b Eng.
bach_brand_conc_no2_bwv1047_andante 4 fl, ob, vn, bc Eng. haydn_str_quartet_op74_no1_menuetto 4 2 vn, va, vc Amer.
beethoven_str_quartet_op18_no1_adagio 4 2 vn, va, vc Amer. sweelinck_miserere_mei 4 SATB Eng.
monteverdi_ave_maris_stella 4 SATB Amer. schubert_an_die_sonne_d439 6 SATB, pf Eng.
mozart_symphony_no1_mvt1_exposition 7 2 ob, 2 hn, 2 vn, va, vc, db Eng.
vivaldi_vn_conc_f_min_rv297_allegro 8 solo vn, 2 vn, va, vc, cb,
hpd Eng.
beethoven_symphony_no3_mvt1_exposition 19 2 fl, 2 ob, 2 cl, 2 bn, 3 hn, Eng.
16
2 tpt, timp, 2 vn, va, vc, db
17
Creation of the Gold Standard of 200 Questions Stage 1
20 scores were collected. 10 questions set on each score. 10 scores: English terminology, 10 scores: American terminology. Mix of question types set against each score. For each question, answers were found in the score. Gradually built up the required question-type distribution. Questions were recorded in C@merata ascii format.
18
Creation of the Gold Standard (cont.) Stage 2 Answers from Stage 1 were carefully checked by two experts Stage 3
Questions in C@merata ascii format automatically converted into C@merata XML format. Stage 4
Test questions automatically extracted from Gold Standard (minus the answers).
19
Evaluation of a Returned Passage Passage is beat correct if it starts at exactly the correct beat (as specified by the divisions value) in the correct start bar and also ends at the correct beat in the end bar. - Useful for applications of results which are themselves automatic. Passage is measure correct if it starts in the bar where the requested feature starts and ends in the bar where the requested feature ends. - Sufficient in many cases for humans to use.
20
Beat Precision (BP) is the number of beat correct passages returned by a system divided by the number of passages (correct or incorrect) returned. Beat Recall (BR) is the number of beat correct passages returned by a system divided by the total number of answer passages known to exist. Measure Precision (MP) is the number of measure correct passages returned by a system divided by the number of passages (correct or incorrect) returned. Measure Recall (MR) is the number of measure correct passages returned by a system divided by the total number of answer passages known to exist.
21
XML Question Format <questions task="camerata" year="2015" task_no="1"
organisation="University of Essex"
group="LAC Group">
<question number="001" music_file="f01.xml" divisions="4">
<text>harmonic perfect fifth</text>
<answer>
</answer>
</question>
</questions>
22
XML Answer Format (for Runs)
<questions task="camerata" year="2015" task_no="1"
runtag="lacg01" organisation="University of Essex"
group="LAC Group">
<question number="001" music_file="f01.xml" divisions="4">
<text>harmonic perfect fifth</text>
<answer>
<passage start_beats="4" start_beat_type="4"
end_beats="4" end_beat_type="4"
start_divisions="4" end_divisions="4"
start_bar="2" start_offset="1"
end_bar="2" end_offset="4" />
<passage start_beats="4" start_beat_type="4"
end_beats="4" end_beat_type="4"
start_divisions="4" end_divisions="4"
start_bar="7" start_offset="5"
end_bar="7" end_offset="8" />
<passage start_beats="4" start_beat_type="4"
end_beats="4" end_beat_type="4"
start_divisions="4" end_divisions="4"
start_bar="9" start_offset=1"
end_bar="9" end_offset=8" />
</answer>
</question>
</questions>
23
Participants - Same as last year! Runtag Leader Affiliation Country
CLAS Stephen Wan CSIRO Australia DMUN Tom Collins De Montfort University England OMDN Donncha Ó Maidín University of Limerick Ireland TNKG Nikhil Kini Mumbai India UNLP Kartik Asooja NUI Galway Ireland
24
2014 Results
Run BP BR BF MP MR MF
CLAS01 0.713 0.904 0.797 0.764 0.967 0.854
DMUN01 0.372 0.712 0.489 0.409 0.784 0.538
DMUN02 0.380 0.748 0.504 0.417 0.820 0.553
DMUN03 0.440 0.868 0.584 0.462 0.910 0.613
LACG01 0.135 0.101 0.116 0.188 0.142 0.162
OMDN01 0.415 0.150 0.220 0.424 0.154 0.226
TCSL01 0.633 0.821 0.715 0.652 0.845 0.736
UNLP01 0.113 0.516 0.185 0.155 0.703 0.254
UNLP02 0.290 0.512 0.370 0.393 0.692 0.501
Maximum 0.713 0.904 0.797 0.764 0.967 0.854
Minimum 0.113 0.150 0.185 0.155 0.154 0.226
Average 0.420 0.654 0.483 0.460 0.734 0.534
25
2015 Results
Run BP BR BF MP MR MF
CLAS01 0.604 0.636 0.620 0.639 0.673 0.656
DMUN01 0.311 0.739 0.438 0.332 0.788 0.467
DMUN02 0.242 0.739 0.365 0.265 0.809 0.399
DMUN03 0.294 0.739 0.421 0.316 0.794 0.452 OMDN01 0.817 0.175 0.288 0.817 0.175 0.288
TNKG01 0.061 0.488 0.108 0.073 0.586 0.129
UNLP01 0.126 0.430 0.195 0.149 0.508 0.230
Maximum 0.817 0.739 0.620 0.817 0.809 0.656
Minimum 0.061 0.175 0.108 0.073 0.175 0.129
Average 0.351 0.564 0.348 0.370 0.619 0.375
26
Summary of Results Based on Beat F-score (BF) 2014 ranking: CLAS01 (0.797), TCSL01 (0.715), DMUN03 (0.584), UNLP02 (0.370), OMDN01 (0.220), LACG01 (0.116) 2015 Ranking: CLAS01 (0.620), DMUN01 (0.438), OMDN01 (0.288), UNLP01 (0.195) and TNKG01 (0.108) Results lower in 2015 because task was much harder
27
2014 Results by Question Type
Type BP BR BF MP MR MF
simple_pitch 0.645 0.736 0.677 0.685 0.787 0.720
simple_length 0.780 0.846 0.810 0.830 0.906 0.864
pitch_and_length 0.662 0.726 0.644 0.719 0.803 0.710
perf_spec 0.339 0.547 0.339 0.350 0.582 0.352
stave_spec 0.408 0.682 0.508 0.432 0.732 0.540
word_spec 0.487 0.771 0.520 0.487 0.771 0.520
followed_by 0.291 0.518 0.278 0.351 0.716 0.355
melodic_interval 0.396 0.417 0.402 0.471 0.501 0.481
harmonic_interval 0.185 0.207 0.188 0.269 0.329 0.281
cadence_spec 0.071 0.141 0.093 0.171 0.297 0.214
triad_spec 0.081 0.125 0.095 0.124 0.171 0.138
texture_spec 0.060 0.109 0.075 0.072 0.141 0.092
28
2015 Results by Question Type
Type BP BR BF MP MR MF
1_melod 0.450 0.764 0.508 0.467 0.801 0.531
n_melod 0.216 0.378 0.249 0.236 0.472 0.276
1_harm 0.261 0.426 0.285 0.289 0.471 0.317
texture 0.000 0.000 0.000 0.143 0.061 0.086 follow 0.172 0.415 0.217 0.247 0.486 0.275
synch 0.193 0.373 0.178 0.235 0.425 0.208
perf qualified 0.359 0.552 0.230 0.362 0.578 0.236
instr qualified 0.426 0.488 0.308 0.440 0.522 0.326
clef qualified 0.329 0.588 0.342 0.339 0.615 0.355
time qualified 0.187 0.476 0.248 0.211 0.544 0.281
key qualified 0.143 0.089 0.110 0.291 0.357 0.300
29
Summary of Question Type Results Based on Beat F-score (BF) 2014 ranking: simple_length (0.810), simple_pitch (0.677), pitch_and_length (0.644), word_spec (0.520), stave_spec (0.508), melodic_interval (0.402), perf_spec (0.339), followed_by (0.278), harmonic_interval (0.188), triad_spec (0.095), cadence_spec (0.093), texture_spec (0.072) 2015 ranking: 1_melod (0.508), 1_harm (0.285), n_melod (0.249), follow (0217), synch (0.178) and texture (0.000). Not all question types are equally easy Big jump from 1_melod to 1_harm - 1_melod generally easier
30
Overall Conclusions We refined our ideas Questions got harder Participant systems got better Technical aspects (data formats, passages, evaluation etc) worked out well Few participants, but they are interested in continuing!
31
The End
32
Example Questions The following slides show some example scores and questions against them.
33
Work: J.S. Bach, Das Wohltemperierte Klavier, Book 1, Prelude No. 2 in C minor BWV 847 Extract:
34
Q: passage in common time A: [4/4,1,1:1-4:4] Q: interval of a melodic sixth A: [4/4,4,1:1-1:2], [4/4,4,1:8-1:9] [4/4,4,1:9-1:10], [4/4,4,2:1-2:2], [4/4,4,2:8-2:9], [4/4,4,2:9-2:10], [4/4,4,3:1-3:2], [4/4,4,3:8-3:9], [4/4,4,3:9-3:10] Q: second A: None, because a second is harmonic by default Q: C followed by Eb A: [4/4,4,1:5-1:6], [4/4,4,1:13-1:14], [4/4,4,4:1-4:2], [4/4,4,4:9-4:10] Q: C followed by Eb in the bass clef A: [4/4,4,4:1-4:2], [4/4,4,4:9-4:10] Q: semiquaver E natural / sixteenth note E natural A: [4/4,4,2:3-2:4]
35
Work: J.S. Bach, Suite No. 3 in C Major for Cello, BWV 1009, Sarabande Extract:
Q: harmonic interval of a minor third A: [3/4,2,208:1-208:1] Q: minor third A: [3/4,2,208:1-208:1] (thirds are harmonic by default) Q: dotted quaver / dotted eighth note A: [3/4,4,206:5-206:7], [3/4,4,207:5-207:7], [3/4,4,208:5-208:7] Q: harmonic perfect fifth A: [3/4,1,206:1-206:1], [3/4,1,207:1-207:1], [3/4,4,208:5-208:7] (in last passage, C and G should be dotted but were not in the original) Q: simultaneous harmonic perfect fifth and harmonic eleventh A: [3/4,4,208:5-208:7] (same point, C and G are assumed dotted in this example)
36
Work: J. Dowland, King of Denmark's Galliard, P 40 Extract:
Q: perfect cadence
A: [3/4,1,3:3-4:3] (We are assuming the cadence continues until the end of the bar) Q: four consecutive quavers / four consecutive quarter notes A: [3/4,2,2:3-2:6] Q: dotted minim in the bass / dotted half note in the bass A: [3/4,1,2:1-2:3] Q: harmonic fourth A: [3/4,1,1:1-1:1], [3/4,1,1:2-1:3], [3/4,1,4:1-4:3] (Note that there are two instances in bar (measure) 1 because the chord is played twice)
37
Work: J. Dowland, Pauana Dulandi, P 86 Extract:
Q: consecutive 5ths in the bass A: [4/4,1,2:1-2:4] Q: semibreve Bb in the treble clef / whole note Bb in the treble clef A: [4/4,1,1:1-1:4], [4/4,1,4:1-4:4] Q: octave followed two bars later by another octave
A: [4/4,1,1:1-3:4] (assumed harmonic) Q: Vc triad A: [4/4,1,2:3-2:3]
38
Work: G.F. Handel, Messiah, HG xlv, "And the Glory of the Lord" Extract:
Q: crotchet rest / quarter note rest A: [3/4,1,1:1-1:1] Q: dotted crotchet followed by three quavers / dotted quarter note followed by three eighth
notes A: [3/4,1,5:1-5:3], [3/4,1,6:1-6:3] Q: four hemidemisemiquavers / four sixty-fourth notes A: [3/4,4,3:9-3:9] Q: three quavers in a row / three eighth notes in a row A: [3/4,2,5:4-5:6], [3/4,2,6:4-6:6]
39
Work: G.F. Handel, Messiah, HG xlv, Overture Extract:
Q: D sharp crotchet / D sharp quarter note
A: [4/4,1,22:1-22:1], [4/4,1,25:1-25:1] Q: D natural quaver / D natural eighth note
A: [4/4,2,23:3-23:3], [4/4,2,24:4-24:4] Q: open VII triad in the first inversion A: [4/4,1,23:3-23:3] Q: melodic fourth in the bass clef A: [4/4,1,24:1-24:2]
40
Work: D. Scarlatti, Keyboard Sonata in D minor, K 1 Extract:
41
Q: eight staccato notes in succession A: [4/4,1,17:1-17:4] Q: augmented melodic fourth A: [4/4,2,19:3-19:4], [4/4,2,20:4-20:5] Q: third
A: [4/4,4,17:13-17:13], [4/4,2,19:2-19:2], [4/4,2,19:4-19:4], [4/4,4,19:9-19:9], [4/4,4,20:9-20:9] (these are assumed to be harmonic and can thus be across parts) Q: a quaver, then a major third / an eighth note, then major third
A: [4/4,2,19:1-19:2], [4/4,2,19:3-19:4] (third assumed to be harmonic and to follow the quaver immediately) Q: change from bass to treble clef A: [4/4,1,17:2p] (recall that this means the clef change is immediately after 17:2)
42
Work: D. Scarlatti, Keyboard Sonata in D major, K 430 Extract:
Q: three fourths
A: [3/8,2,56:1-56:3] (assumed consecutive) Q: treble clef F natural A: [3/8,2,62:2-62:2] Q: melodic octave
A: [3/8,4,57:1-57:2], [3/8,4,57:2-57:6] (melodic must be stated, otherwise it is harmonic) Q: harmonic 5th followed by harmonic 4th A: [3/8,2,58:1-58:2], [3/8,2,62:1-62:2]
43
Architecture of Baseline System We have built a very basic system to perform the task: 1. Analyse the input xml file and extract questions and score files 2. For each question:
• Parse question using Stanford Parser • Determine question type (similar to normal QA) • Parse the score file using Music21 • Based on question type, search for answer passages in score 3. Write out answers to output xml file
44
Linguistic Observations Queries are noun phrases Head noun group is the main feature (e.g. F#) PP modifiers qualify this (e.g. in the bass clef) A lot of terminology is used (e.g. semiquaver, soprano part, forte bar) Search based on query classification will work at least for simple examples.
45
Summary We are trying to find ways to link natural language descriptions of music to musical scores. Starting with the C@merata task, we are working with very simple tasks to develop the technology. Then we will progress to more complex tasks which are of genuine interest to musicologists. In due course, we aim to tie passages in musicological texts such as in Grove Online to the corresponding music scores.
46
Music Extracts symphony opening with a horn call over shimering strings Bruckner 4th Symphony http://www.youtube.com/watch?v=J8t1TzN0RRY Start: 0:00 End: 0:44 symphony closing with six unison chords Sibelius 5th Symphony http://www.youtube.com/watch?v=nkzrSZKA4cM Start: 9:36 End: 9:56 Nielsen 4th Symphony, Finale (shows two timpanists well) http://www.youtube.com/watch?v=yXDe1hj4HBo Start: 32:10 End: 32:48
47
Extracts Not Used symphony opening with two octave descending arpeggio Beethoven 9th Symphony string quartet opening including double stopping on all four instruments Beethoven op 127 symphony featuring a battle between two timpanists Nielsen 4th Symphony, Finale (not so good) http://www.youtube.com/watch?v=sD9I-UiYfW8 Start: 6:30 End: 7:08
48
Grove’s Dictionary of Music and Musicians • 1879-1889 George Grove, civil engineer, music administrator, writer and then Director of
Royal College of Music, wrote A Dictionary of Music and Musicians in four volumes. • 1904-910 Fuller Maitland edited the second edition - Grove's Dictionary of Music and
Musicians - in five volumes. • 1927 Henry Colles edited the third edition - in five volumes. • 1940 Henry Colles edited the fourth edition - in seven volumes. • 1954 Eric Blom edited the fifth edition - in nine volumes.
49
• 1980 Stanlie Sadie edited the sixth edition - The New Grove Dictionary of Music and
Musicians - in twenty volumes. Contained 22,500 articles and 16,500 biographies. • 2001 Stanlie Sadie edited the seventh edition - in 29 volumes. This was also available
online. • 2009 Deane Root was appointed editor of Grove Music Online. By this time it contained
more than 50,000 articles. Grove is now considered to be the most comprehensive and scholarly source of information on Western Classical Art Music which exists. It is used daily by musicians and musicologists worldwide.
50
What might our queries be like in future? We will show some examples:
• A natural language phrase describes a musical feature.
• A regular expression or other pattern could not capture it.
51
Some Examples Q: 'horn call over shimmering strings'
52
Some Examples Q: 'horn call over shimmering strings' A: Anton Bruckner, Symphony No. 4 "Romantic"
53
Some Examples Q: 'horn call over shimmering strings' A: Anton Bruckner, Symphony No. 4 "Romantic" Q: 'movement ending with six stacatto chords for full orchestra'
54
Some Examples Q: 'horn call over shimmering strings' A: Anton Bruckner, Symphony No. 4 "Romantic" Q: 'movement ending with six stacatto chords for full orchestra' A: Jean Sibelius, Symphony No. 5
55
Some Examples Q: 'horn call over shimmering strings' A: Anton Bruckner, Symphony No. 4 "Romantic" Q: 'movement ending with six stacatto chords for full orchestra' A: Jean Sibelius, Symphony No. 5 Q: 'symphony featuring a battle between two timpanists'
56
Some Examples Q: 'horn call over shimmering strings' A: Anton Bruckner, Symphony No. 4 "Romantic" Q: 'movement ending with six stacatto chords for full orchestra' A: Jean Sibelius, Symphony No. 5 Q: 'symphony featuring a battle between two timpanists' A: Carl Nielsen, Symphony No. 4 "The Inextinguishable"
57
What are the Characteristics of these Queries? Not very long Not that specific ('over', 'featuring') Use musical terms ('horn', 'strings', 'stacatto', 'chords', 'full orchestra', 'symphony', 'timpani') Also use non-musical terms, interpreted in a musical way ('shimmering', 'battle')
58
What are the Characteristics of these Queries? Not very long Not that specific ('over', 'featuring') Use musical terms ('horn', 'strings', 'stacatto', 'chords', 'full orchestra', 'symphony', 'timpani') Also use non-musical terms, interpreted in a musical way ('shimmering', 'battle') Nevertheless, experts can answer them!
59
Table 4. Results for simple_pitch Questions
Run BP BR BF MP MR MF
CLAS01 0.959 0.972 0.965 0.982 0.995 0.988
DMUN01 0.717 0.674 0.695 0.790 0.743 0.766
DMUN02 0.729 0.729 0.729 0.798 0.798 0.798
DMUN03 0.955 0.972 0.963 0.968 0.986 0.977
LACG01 0.000 0.000 0 0.200 0.028 0.049
OMDN01 0.000 0.000 0.000 0.000 0.000 0.000
TCSL01 0.959 0.963 0.961 0.982 0.986 0.984
UNLP01 0.422 0.789 0.550 0.478 0.894 0.623
UNLP02 0.422 0.789 0.550 0.478 0.894 0.623
Maximum 0.959 0.972 0.965 0.982 0.995 0.988
Minimum 0.000 0.000 0.000 0.000 0.000 0.000
Average 0.645 0.736 0.677 0.685 0.787 0.720
60
Table 5. Results for simple_length Questions
Run BP BR BF MP MR MF
CLAS01 0.904 0.988 0.944 0.915 1.000 0.956
DMUN01 0.858 0.852 0.855 0.879 0.874 0.876
DMUN02 0.863 0.889 0.876 0.884 0.911 0.897
DMUN03 0.955 0.985 0.970 0.967 0.997 0.982
LACG01 0.529 0.197 0.287 0.620 0.231 0.337
OMDN01 0.408 0.471 0.437 0.419 0.483 0.449
TCSL01 0.979 0.988 0.983 0.991 1.000 0.995
UNLP01 0.636 0.797 0.707 0.791 0.991 0.880
UNLP02 0.636 0.797 0.707 0.791 0.991 0.880
Maximum 0.979 0.988 0.983 0.991 1.000 0.995
Minimum 0.408 0.471 0.437 0.419 0.483 0.449
Average 0.780 0.846 0.810 0.830 0.906 0.864
61
Table 6. Results for pitch_and_length Questions
Run BP BR BF MP MR MF
CLAS01 0.860 0.937 0.897 0.895 0.975 0.933
DMUN01 0.653 0.785 0.713 0.721 0.867 0.787
DMUN02 0.663 0.823 0.734 0.730 0.905 0.808
DMUN03 0.760 0.943 0.842 0.770 0.956 0.853
LACG01 0.157 0.196 0.174 0.172 0.215 0.191
OMDN01 0.714 0.032 0.061 0.714 0.032 0.061
TCSL01 0.723 0.892 0.799 0.754 0.930 0.833
UNLP01 0.460 0.696 0.554 0.582 0.880 0.701
UNLP02 0.460 0.696 0.554 0.582 0.880 0.701
Maximum 0.86 0.943 0.897 0.895 0.975 0.933
Minimum 0.460 0.032 0.061 0.582 0.032 0.061
Average 0.662 0.726 0.644 0.719 0.803 0.710
62
Table 7. Results for perf_spec Questions
Run BP BR BF MP MR MF
CLAS01 1.000 0.862 0.926 1.000 0.862 0.926
DMUN01 0.407 0.379 0.393 0.444 0.414 0.428
DMUN02 0.407 0.379 0.393 0.444 0.414 0.428
DMUN03 0.741 0.690 0.715 0.741 0.690 0.715
LACG01 0.000 0.000 0.000 0.000 0.000 0.000
OMDN01 0.000 0.000 0.000 0.000 0.000 0.000
TCSL01 0.066 0.897 0.123 0.066 0.897 0.123
UNLP01 0.045 0.586 0.084 0.053 0.690 0.098
UNLP02 0.045 0.586 0.084 0.053 0.690 0.098
Maximum 1.000 0.897 0.926 1.000 0.897 0.926
Minimum 0.000 0.000 0.000 0.000 0.000 0.000
Average 0.339 0.547 0.339 0.350 0.582 0.352
63
Table 8. Results for stave_spec Questions
Run BP BR BF MP MR MF
CLAS01 0.568 1.000 0.724 0.568 1.000 0.724
DMUN01 0.534 0.840 0.653 0.568 0.893 0.694
DMUN02 0.534 0.840 0.653 0.568 0.893 0.694
DMUN03 0.619 0.973 0.757 0.619 0.973 0.757
LACG01 0.165 0.240 0.196 0.174 0.253 0.206
OMDN01 0.000 0.000 0.000 0.000 0.000 0.000
TCSL01 0.661 0.987 0.792 0.661 0.987 0.792
UNLP01 0.173 0.440 0.248 0.230 0.587 0.331
UNLP02 0.173 0.373 0.236 0.241 0.520 0.329
Maximum 0.661 1.000 0.792 0.661 1.000 0.792
Minimum 0.000 0.000 0.000 0.000 0.000 0.000
Average 0.408 0.682 0.508 0.432 0.732 0.540
64
Table 9. Results for word_spec Questions
Run BP BR BF MP MR MF
CLAS01 1.000 1.000 1.000 1.000 1.000 1.000
DMUN01 0.750 0.750 0.750 0.750 0.750 0.750
DMUN02 0.750 0.750 0.750 0.750 0.750 0.750
DMUN03 1.000 1.000 1.000 1.000 1.000 1.000
LACG01 0.044 0.250 0.075 0.058 0.333 0.099
OMDN01 0.000 0.000 0.000 0.000 0.000 0.000
TCSL01 0.261 1.000 0.414 0.261 1.000 0.414
UNLP01 0.067 0.833 0.124 0.067 0.833 0.124
UNLP02 0.067 0.833 0.124 0.067 0.833 0.124
Maximum 1.000 1.000 1.000 1.000 1.000 1.000
Minimum 0.000 0.000 0.000 0.000 0.000 0.000
Average 0.487 0.771 0.520 0.487 0.771 0.520
65
Table 10. Results for followed_by Questions
Run BP BR BF MP MR MF
CLAS01 0.748 0.859 0.800 0.830 0.953 0.887
DMUN01 0.090 0.797 0.162 0.093 0.820 0.167
DMUN02 0.092 0.820 0.165 0.094 0.844 0.169
DMUN03 0.094 0.844 0.169 0.096 0.859 0.173
LACG01 0.003 0.008 0.004 0.068 0.156 0.095
OMDN01 0.567 0.133 0.215 0.567 0.133 0.215
TCSL01 0.733 0.688 0.710 0.842 0.789 0.815
UNLP01 0.000 0.000 0.000 0.025 0.695 0.048
UNLP02 0.000 0.000 0.000 0.260 0.633 0.369
Maximum 0.748 0.859 0.800 0.842 0.953 0.887
Minimum 0.000 0.000 0.000 0.025 0.133 0.048
Average 0.291 0.518 0.278 0.351 0.716 0.355
66
Table 11. Results for melodic_interval Questions
Run BP BR BF MP MR MF
CLAS01 0.660 0.837 0.738 0.699 0.886 0.781
DMUN01 0.528 0.545 0.536 0.724 0.748 0.736
DMUN02 0.521 0.610 0.562 0.701 0.821 0.756
DMUN03 0.562 0.659 0.607 0.736 0.862 0.794
LACG01 0.000 0.000 0.000 0.158 0.024 0.042
OMDN01 0.000 0.000 0.000 0.000 0.000 0.000
TCSL01 0.894 0.683 0.774 0.904 0.691 0.783
UNLP01 0.000 0.000 0.000 0.000 0.000 0.000
UNLP02 0.000 0.000 0.000 0.000 0.000 0.000
Maximum 0.894 0.837 0.774 0.904 0.886 0.794
Minimum 0.000 0.000 0.000 0.000 0.000 0.000
Average 0.396 0.417 0.402 0.471 0.501 0.481
67
Table 12. Results for harmonic_interval Questions
Run BP BR BF MP MR MF
CLAS01 0.158 0.429 0.231 0.353 0.957 0.516
DMUN01 0.415 0.386 0.400 0.585 0.543 0.563
DMUN02 0.415 0.386 0.400 0.585 0.543 0.563
DMUN03 0.492 0.457 0.474 0.631 0.586 0.608
LACG01 0.091 0.014 0.024 0.273 0.043 0.074
OMDN01 0.000 0.000 0.000 0.000 0.000 0.000
TCSL01 0.000 0.000 0.000 0.000 0.000 0.000
UNLP01 0.000 0.000 0.000 0.000 0.000 0.000
UNLP02 0.000 0.000 0.000 0.000 0.000 0.000
Maximum 0.492 0.457 0.474 0.631 0.957 0.608
Minimum 0.000 0.000 0.000 0.000 0.000 0.000
Average 0.185 0.207 0.188 0.269 0.329 0.281
68
Table 13. Results for cadence_spec Questions
Run BP BR BF MP MR MF
CLAS01 0.238 0.625 0.345 0.286 0.750 0.414
DMUN01 0.083 0.125 0.100 0.333 0.500 0.400
DMUN02 0.083 0.125 0.100 0.333 0.500 0.400
DMUN03 0.167 0.250 0.200 0.417 0.625 0.500
LACG01 0.000 0.000 0.000 0.200 0.125 0.154
OMDN01 0.000 0.000 0.000 0.000 0.000 0.000
TCSL01 0.000 0.000 0.000 0.000 0.000 0.000
UNLP01 0.000 0.000 0.000 0.000 0.000 0.000
UNLP02 0.000 0.000 0.000 0.000 0.000 0.000
Maximum 0.238 0.625 0.345 0.417 0.75 0.500
Minimum 0.000 0.000 0.000 0.000 0.000 0.000
Average 0.071 0.141 0.093 0.171 0.297 0.214
69
Table 14. Results for triad_spec Questions
Run BP BR BF MP MR MF
CLAS01 0.348 0.727 0.471 0.391 0.818 0.529
DMUN01 0.100 0.091 0.095 0.200 0.182 0.191
DMUN02 0.100 0.091 0.095 0.200 0.182 0.191
DMUN03 0.100 0.091 0.095 0.200 0.182 0.191
LACG01 0.000 0.000 0.000 0.000 0.000 0.000
OMDN01 0.000 0.000 0.000 0.000 0.000 0.000
TCSL01 0.000 0.000 0.000 0.000 0.000 0.000
UNLP01 0.000 0.000 0.000 0.000 0.000 0.000
UNLP02 0.000 0.000 0.000 0.000 0.000 0.000
Maximum 0.348 0.727 0.471 0.391 0.818 0.529
Minimum 0.000 0.000 0.000 0.000 0.000 0.000
Average 0.081 0.125 0.095 0.124 0.171 0.138
70
Table 15. Results for texture_spec Questions
Run BP BR BF MP MR MF
CLAS01 0.182 0.500 0.267 0.273 0.750 0.400
DMUN01 0.100 0.125 0.111 0.100 0.125 0.111
DMUN02 0.100 0.125 0.111 0.100 0.125 0.111
DMUN03 0.100 0.125 0.111 0.100 0.125 0.111
LACG01 0.000 0.000 0.000 0.000 0.000 0.000
OMDN01 0.000 0.000 0.000 0.000 0.000 0.000
TCSL01 0.000 0.000 0.000 0.000 0.000 0.000
UNLP01 0.000 0.00 0.000 0.000 0.000 0.000
UNLP02 0.000 0.000 0.000 0.000 0.000 0.000
Maximum 0.182 0.500 0.267 0.273 0.750 0.400
Minimum 0.000 0.000 0.000 0.000 0.000 0.000
Average 0.060 0.109 0.075 0.072 0.141 0.092