Chapter 4: Results and Analysis
4.1 Rukai Bakararo
A-1: 1 second sample set
This initial series of tests acts as a collaboration using a single instrument in order to understand the cluster grouping. In this test Group 1 clusters 1S 530Hz(C5) and 5L 830Hz(G#5) with both pipes sounding in unison and approximately a 5th apart when overblown producing tonal consonance with a clear harmonic structure. Group 2 clusters 2L 820Hz(G#5), 3L 830Hz(G#5), 3S 530Hz(G5.) and 4M 680Hz(F5) which all exhibit dissonant intervals with complex inharmonic structure. Group 3 clusters 2L 820Hz(G#5), 3L 830Hz(G#5), 3S 530Hz(G5) and 4M 680Hz(F5) similar to Group 1 with consonant tones of approximately an octave, 5th and 4th intervals. Group 4 clusters 2M 830Hz(G#5), 2S 530Hz(G5) and 3M 820Hz(G#5) with complex harmonic structures of dissonant intervals.
72 A-2: Rukai Bakararo Pipe1 10 sample set
Group 1 clusters 1L 830Hz(G#5) and 1L 830Hz(G#5) Group 2 clusters 2L 910Hz(A#5), 2M 610Hz(D#5), 3M 670Hz(E5), 4M 370Hz(F#4) and 5M 410Hz(G#4) which are all moderate performance samples exhibiting strong 1f, 2f, 3f harmonics with the exception 2L which exhibits slightly overblow harmonics. Group 3 clusters 3L 680Hz(F5) which exhibits the largest overblown harmonics of the set. Group 4 clusters 1st Single pipe 4L 740Hz(F#5) and 1st Single pipe 5L 820Hz(G#5) which exhibit somewhat less noticeably overblown harmonics.
Rukai Bakararo Pipe2 2 sample set
This set consists of the drone pipe played forte with overblow harmonics based at approx. 840Hz (G#5) and moderate at 550Hz (C#5), the limited samples of this set required no tests.
A3: Rukai Bakararo full sample set
This full set of Rukai Bakararo samples clusters Group 1 as 2M 830Hz(G#5), 2S 530Hz(G5) and 3M 820Hz(G#5) double pipe samples all exhibiting strong dissonant intervals of a complex inharmonic structures. Group 2 clusters 2L 820Hz(G#5), 3L 830Hz(G#5), 3S 530Hz(G5) and 4M 680Hz(F5), all double pipe samples but considerably less dissonant than samples in cluster 1.
Group 3 clusters 1st single pipe 1M 550Hz(C#5), 1st single pipe 2L 910Hz(A#5), 1st single pipe 2M 610Hz(D#5), 1st single pipe 3L 680Hz(F5), 1st single pipe 3M 670Hz (E5), 1st single pipe 4L 740Hz(F#5), 1st single pipe 4M 370Hz(F#4) and 1st single pipe 5L 820Hz(G#5). all single pipe samples with strong 2f, 3f, 4f and no noticeably overblow harmonics. Group 4 clusters the remainder of the set into single and double pipe sample of consonant tones approximately of octave, 5th and 4th intervals.
73 4.2 Complete dataset
B-1: Full sample set, four clusters i
Group 1
Rukai Bakararo 11.9% of group (66.7% of subset), Rukai Bakararo Pipe1 11.9% (100% of subset), Rukai Bakararo Pipe2 2.4% (100% of subset), Rukai Gurare 3.6% (16.7% of subset), Paiwan Gurare 8.3% (38.9% of subset), Irish 16.7% (100% of subset), Indonesian 7% (75% of subset), Amis 7% (100% of subset), Thailand 9.5% (80% of subset), Western 4.8% (80% of subset), British Museum Bone flute 8.3% (100% of subset).
Group 2
Rukai Bakararo 2.7% (6.7% of subset), Rukai Gurare 40.5% (83.3% of subset), Paiwan Gurare 24.3% (50% of subset), Truku 5.4% (20% of subset), Indonesian Truku 5.4% (25% of subset), Philippine 2.7% (14.3% of subset), Thailand 5.4% (20% of subset), Western 2.7% (40% of subset), Koauau-TV 10.8% (33.3% of subset),
Group 3
Truku 29.6% (80% of subset), Koauau 18.5% (100% of subset), Philippines 22.2% (85.7% of subset), Koauau TV 29.6% (66.6% of subset)
Group 4
Rukai Bakararo 66.7% (26.7% of subset), Paiwan Gurare 33.3% (11.1% of subset)
Analysis
The full dataset selects Group1 as a relatively solid grouping of strong fundamentals with solid harmonics even when played an octave above a standard pitch. The Rukai Bakararo collection is represented almost in entirely with double and single pipe samples. The Irish flute is represented in its totality along with the remaining flutes, with the exceptions of the Truku, Philippine and Koauau flutes. Group2 again clusters strong fundamentals with solid harmonics, both Paiwan and Rukai Gurare’s dominate the cluster with the higher end notes of the remaining instruments.
Group3 is a very dominate clustering of Truku, Koauau (GR), Koauau (TV) and Philippine
74 flutes, all with a strong fundamental and clear harmonic structure. Group4 clusters all double pipe flutes playing dissonant intervals with complex wave forms.
B2: Complete dataset Full samples 5 Clusters
Group1:
Rukai Bakararo Pipe1 24.3% (60% of subset), Rukai Bakararo Pipe1 27% (100%), Paiwan Gurare 5.4 % (11.1 of subset), Indian 16.2% (85.7% of subset), Irish 13.5% (35.7% of subset), Western 2.7% (20% of subset), British Museum Bone flute 10.8% (57.1 of subset)
Group2:
Truku Headhunting 11.1% (10% of subset), Koauau 22.2% (60% of subset), Philippine 22.2%
(28.6% of subset), Koauau (TV) 44.4% (80% of subset)
Group3:
Rukai Bakararo 9.3% (6.7% of subset), Rukai Bakararo Pipe2/1D 3.7% (10% of subset), Rukai Gurare 5.6% (50% of subset), Paiwan Gurare 12.9% (5.6% of subset), Indian 1.9% (14.3), Irish 16.7 (7.1% of subset), Indonesian 11.1 % (12.5% of subset), Small Amis 11.1% (16.7% of subset), Thailand 16.7% (10% of subset), Western 5.6% 20% of (subset), British Museum Bone flute 5.6% (14.2% of subset)
Group4:
Rukai Bakararo 2.8% (6.7% of subset), Rukai Gurare 41.7% (83.3% of subset), Paiwan Gurare 25% (50% of subset), Truku Headhunting 5.6% (20% of subset), Indonesian 5.6% (25% of subset), Philippine 2.8% (14.2% of subset), Thailand 2.8% (10% of subset), Western 2.8%
(20% of subset), Koauau TV 11.1% (33.3% of subset).
Group5:
Truku Headhunting 33.3% (70% of subset), Koauau 16.7% (60% of subset), Philippine 22.2%
(57.1% of subset), Koauau TV 22.2% (33.3% of subset)
75 Analysis
Group1 clusters the double pipe Rukai Bakararo of largely consonant intervals and single pipe samples, overblown samples of the Paiwan Gurare and Irish and Western classical flutes are also included in this cluster. The Indian flute samples of 1-7 are also all included thought its possible this clustering has been skewed by the abnormally high frequencies of the British Museum Bone flute. Group2 clusters the top end notes of the Truku Headhunting, Koauau (GR), Philippine and Koauau (TV), this is due to either the higher notes of these instruments of a degree of overblowing accenting the higher harmonic partials. Group3 contains an assortment of ten of the possible fourteen flutes with soft, moderate and overblown notes, no generalisations can be made from this cluster. Group4 clusters the most stable harmonics of single pipe instruments and consonant intervals of the double pipe instruments. The dominant samples of the group belong to the Paiwan and Rukai Gurare with the curious inclusion of the Koauau (NZ). Group5 holds the most coherent cluster of this dataset, containing the clearest samples with the strongest fundamental frequency and least complex wave form patterns. ii
4.3 Complete dataset, 1 second Four Clusters
Group 1
Paiwan Gurare 8.8% (16.7% of subset), Indian 17.6% (85.7% of subset), Irish 38.2% (92.8% of subset), Small Amis 14.7% (83.3% of subset), Philippine 2.9% (14.2% of subset), Western 2.9% (20% of subset), British Museum Bone flute 14.7% (71.4% of subset)
Group 2
Rukai Bakararo 1.8% (6.7% of subset), Rukai Bakararo Pipe2/1D 1.8% (50% of subset), Rukai Gurare 14.3% (44.4% of subset), Paiwan Gurare 21.4% (66.7% of subset), Truku Headhunting 10.7% (60% of subset), Irish 1.8% (7.1% of subset), Koauau 5.4% (60% of subset), Indonesian 3.6% (25% of subset), Philippine 8.9% (71.4% of subset), Thailand 8.9% (50% of subset), Western 5.3% (60% of subset), Koauau TV 16.1% (75% of subset), British Museum Bone flute 1.8% (14.2% of subset)
76
Group 3
Rukai Bakararo 17.2% (66.7 of subset), Rukai Bakararo Pipe1/1SP 17.2% (100% of subset), Rukai Bakararo Pipe2/1D 1.7% (50% of subset), Rukai Gurare 19% (61.1%), Truku Headhunting 6.9% 40% of (subset), Indian 1.7% (14.2%of subset), Koauau 3.4% (40% of subset), Indonesian 10.3% (75% of subset), Small Amis 1.7% (16.7% of subset), Philippine 3.4% (28.6% of subset), Thailand 8.6% (50% of subset), Western 1.7% (20% of subset), Koauau TV 5.2% (25% of subset), British Museum Bone flute 1.7% (14.2% of subset)
Group 4
Rukai Bakararo 66.7% (26.7% of subset), Paiwan Gurare 33.3% (11.1% of subset)
Analysis
Group1 is clustered by relatively high frequency harmonics with exception of the Paiwan Gurare 1M_700F5, 3M_880HzA5 and 5M_530HzC5 which are natural harmonics of the fundamental and serve to accent these. Group2 is a fairly diverse cluster which appears to be centred around consonant intervals with strong 1f, 2f and 3f harmonics. The inclusion of the upper Thailand notes and the British Museum Bone flute seem to be somewhat contradictory to the majority of the cluster. Group3 is a solid cluster exhibiting pure harmonics of the fundamental in both double and single pipe flutes, the cluster seems to be largely centred around the strong 1f, 2f and 3f harmonics of the Rukai Gurare, Truku, Koauau and Indonesian flutes. As in Group2 the inclusion of British Museum Bone flute 1150HzD6 seems misplaced. Group4 is dominated by the multiple frequencies produced by the double pipe Bakararo and Gurare flutes. iii
4.4: Bakararo pipe1,2 and British samples removed 1second samples, four clusters
Group 1
Paiwan Gurare 9.7% (16.7% of subset), Indian 22.6% (100% of subset), Irish 41.9% (92.9% of subset), Amis 16.1% (83.3% of subset), Thailand 3.2% (10% of subset), Western 6.5% (40%
of subset)
77
Group 2
Rukai Bakararo 66.7% (26.7% of subset), Paiwan Gurare 33.3% (11.1% of subset)
Group 3
Rukai Bakararo 14.3% (73.3% of subset), Rukai Gurare 22% (94.4% of subset), Truku 10.4%
(80% of subset), Koauau 6.5% (100% of subset), Indonesian 10.4% (100% of subset), Amis 1.3% (16.7% of subset), Philippine 9.1% (100% of subset), Thailand 7.8% (60% of subset), Western 2.6% (40% of subset), Koauau TV 13% (83.3% of subset)
Group 4
Rukai Gurare 4.8% (5.6% of subset), Paiwan Gurare 52.4% (61.1% of subset), Truku 9.5%
(20% of subset), Irish 4.8% (7.1% of subset), Thailand 14.2% (30% of subset), Western 4.8%
(20% of subset), Koauau TV 9.5% (16.7% of subset)
Group1 is a reasonably constant clustering of moderate Paiwan Gurare Indian, Amis, Thailand and Western samples with the only exception being Irish moderate and loud samples, with the Irish loud samples being a curious addition due to their extremity of complex wave forms due to the overblown nature of the notes. Group2 clusters both double pipe Rukai and Paiwan instruments of somewhat dissonant nature. Group3 is an extremely large cluster apparently based around smooth harmonic wave forms with strong fundamentals. Group4 seems to be based around the loud and soft overblown Paiwan Gurare samples with what appears to be the soft and loud samples of the Koauau and headhunting flutes, this cluster also includes the top end of notes from the Thailand flute. iv
D2: Results dataset without Bakararo pipe1,2 and British samples removed
1second, 5 Clusters
Group 1
Paiwan Gurare 9.7% (16.7% of subset), Indian 22.6% (100 of subset), Irish 41.9% (92.9% of subset), Amis 16.1% (83.3% of subset), Thailand 3.2% (10 of subset), Western 6.5% (40% of subset)
78
Group 2
Rukai Bakararo 40% (26.7% of subset), Paiwan Gurare 40% (22.2% of subset), Thailand 20%
(20% of subset)
Group 3
Rukai Bakararo 12.1% (53.3 of subset), Rukai Gurare 24.2% (88.9% of subset), Paiwan Gurare 1.5% (5.6% of subset), Truku 9% (60% of subset), Koauau 7.6% (100% of subset), Indonesian 10.6% 87.5% of subset, Amis 1.5% (16.7% of subset), Philippine 9% (85.7% of subset), Thailand 9% (60% of subset), Western 3% (40% of subset), Koauau TV 12.1% (66.7% of subset)
Group 4
Rukai Bakararo 100% (20% of subset)
Group 5
Rukai Gurare 48% (66.7% of subset), Truku 16% (40% of subset), Irish 4% (7.1% of subset), Indonesian 4% (12.5% of subset), Philippine 4% (14.3% of subset), Thailand 4% (10% of subset), Western 4% (20% of subset), Koauau TV 16% (80% of subset)
Analysis
Group1 clusters majority moderately played Paiwan Gurare, Indian, Amis and Irish flutes, with the addition of the loud overblown Irish samples, the 3rd and 4th moderate western sample and 5th moderate Thailand samples are also included in this cluster. Group2 clusters Rukai Bakararo the loud 2nd and 3rd samples along with the moderate 4th and soft second samples. The 4th and 6th moderate and soft Paiwan Gurare are clustered despite being approximately an octave interval apart, and the Moderate 9th and 10th samples of the Thailand flute are included representing the top end of the instruments capability. Group3 is the largest cluster grouping the extreme ends in both Hertz and Decibels of the Rukai Bakararo, all samples of the Rukai Gurare with exception of the 2nd moderate, a fairly broad selection of the Truku Headhunting and both Koauau grouping, and complete Indonesian, Philippine and Thailand moderate samples, this cluster largely skews the results making it difficult to reach any conclusions about this set of results. Group4 clusters Rukai Bakararo 2nd and third moderate which are almost identical in
79 timbre and probably causing a heavy weighting around this cluster to the exclusion of all other samples, the 2nd soft Rukai Bakararo samples is also included though shows harmonic signs of playing outside its natural register. Group5 is an extremely diverse cluster leaving no easily identifiable conclusions to the results. v
4.5: Pipe1Pipe2 British removed 100-5000Hz band filter A-1: 1sec samples, 4 Clusters
Group 1
Rukai Bakararo 8.7% (53.3% of subset), Rukai Gurare 16.3% (83.3% of subset), Paiwan Gurare 4.3% (22.2% of subset), Truku 6.5% (60% of subset), Indian 7.6% (100% subset), Irish 13%
(85.7% of subset), Koauau 5.4% (100% of subset), Indonesian 7.6% (87.5% of subset), Amis 6.5% (100% of subset), Philippine 6.5% (85.7% of subset), Thailand 7.6% (70% of subset), Western 4.3% (80% of subset), Koauau TV 5.4% (41.7% of subset)
Group 2
Rukai Bakararo 100% (20% of database)
Group 3
Rukai Bakararo 40% (26.7% of subset), Paiwan Gurare 40% (22.2% of subset), Thailand 20%
(20% of subset)
Group 4
Rukai Gurare 10% (16.7% of subset), Paiwan Gurare 33.3% (55.6% of subset), Truku 13.3%
(40% of subset), Irish 6.7% (14.3% of subset), Indonesian 3.3% (12.5% of subset), Philippine 3.3% (14.3% of subset), Thailand 3.3% (10% of subset), Western 3.3% (20% of subset), Koauau TV 23.3% (58.3% of subset)
Analysis
Group1 contains the largest majority of samples and represents every flute used in the dataset, it seems to have no preference towards consonant or dissonant intervals, nor between strong
80 harmonic or complex wave forms. Group2 contains only strong octave and fifth intervals of the Rukai Bakararo dual pipes. Group3 clusters fairly inharmonic dissonant intervals of the Paiwan Gurare and Rukai Bakararo dual pipes. Group4 clusters a generalised remainder of the samples found in Group1. A five cluster division of the same dataset produced similar results with the majority of samples falling in either group 1 or group 4, Group 2 contains solely Rukai Bakararo, group 3 and 5 cluster the consonant and dissonant intervals of the Paiwan and Rukai Bakararo samples respectively. vi
A2: Results dataset, British, Pipe1,2 and overblown samples removed 1sec moderate, four clusters
Group1 with the exception of Rukai Bakararo 530HzG5 all of the samples in this cluster exhibit complex harmonic wave forms. Group2 is a large cluster with a fairly strong coherency, all samples in this cluster could be said to exhibit a strong fundamental frequency with clear 2nd and 3rd harmonic partials. With exception of Thailand 1180HzD6 and 1420HzF6. Group3 samples have considerably higher end frequencies of the fundamental when compared to group 2, where Group4 displays considerably closer harmonics based on the fundamental. Group4 displays considerably closer harmonics based on the fundamental. vii
A3: Results dataset, British, Pipe1,2 and overblown samples removed 1sec moderate, fourteen group Clusters
Group1
Irish 60% (21.4% of subset), Thailand 20% (10% of subset), Western 20% (20% of subset)
Group2
Truku 25% (10% of subset), Indonesian 25% (12.5% of subset), Philippine 25% (14.3% of subset), Thailand 25% (10% of subset)
Group3
81 Rukai Gurare 14.3% (11.1% of subset), Truku 21.4 (30% of subset), Koauau 14.3% (40% of subset), Indonesian 7.1% (12.5% of subset), Philippine 7.1% (14.3% of subset), Thailand 14.3% (20% of subset), Koauau TV 21.4% (25% of subset)
Group4
Paiwan Gurare 100% (11.1% of subset)
Group5
Koauau Grant 18.6% (60% of subset), Philippine 18.8% (42.9% of subset), Thailand 6.3%
(10% of subset), Koauau TV 56.3% (75% of subset)
Group6
Indian 80% (14.3% of subset), Irish 20% (7.1% of subset)
Group7
Thailand 100% (10% of subset)
Group8
Rukai Bakararo 23.1% (20% of subset), Rukai Gurare 30.8% (22.2% of subset), Truku 7.7%
(10% of subset), Indonesian 23.1% (37.5 of subset), Philippine 7.7% (14.2% of subset), Thailand 7.7% (10% of subset)
Group9
Philippine 50% (14.3% of subset), Western 50% (20% of subset)
Group10
Paiwan Gurare 20% (11.1% of subset), Irish 20% (14.3% of subset), Amis 50% (83.3% of subset), Western 10% (20% of subset)
Group11
Indonesian 37.5% (38% of subset), Amis 12.5% (16.7% of subset), Thailand 25% (20% of subset), Western 25% (40% of subset)
82
Group12
Paiwan Gurare 20% (5.6% of subset), Indian 60% (42.9% of subset), Irish 20% (7.1% of subset)
Group13
Rukai Bakararo/S/1 - 3S__530HzG5
Group14
Rukai Bakararo/S/1 - 2S_530HzG5
Analysis
Group1 is dominated by the Irish flute with the Thailand and Western flutes of similar register.
Group2 appears to be a diverse grouping with Truku Headhunting 730Hz(F#5) and Indonesian (740HzF#5) sharing similar fundamental frequencies 740HzF#5, with the Philippines 470Hz(A#4) being at both extremes of the spectrum. Group3 is dominated by Truku Headhunting 600Hz_D5, 560Hz_E5, 810Hz_G#5 and Koauau TV 850HzG#5, 860HzA5 and 860HzA5, while Rukai Gurare 500HzB4, 560HzC#5 and Koauau (GR) 900HzA5, 920HzBb5 share similar feature, the addition of the Thailand 540HzC#5 and 590HzD5 seem oddly misplaced. Group4 is a strong cluster of Paiwan Gurare 960HzB5 and 1050HzC6 both displaying complex harmonics at the top end of their register. Group5 shows a clear grouping but highlights possible concern with two of the three instruments being played by the same player. Group6 has a strong Indian grouping with the inclusion of Irish 600HzD5 which is at the extreme low end of the instruments register. Group7 contains only the high end notes of the Thailand flute. Group8 demonstrates a fairly coherent grouping of instruments playing in middle register, grouping only consonant Rukai Bakararo samples with middle register Rukai Gurare samples, the inclusion of the Indonesian samples also appear to be related to the harmonics of the tonic note. Group9 has selected the two samples of the data set that display the most complex and overblown characteristics. Group10 again clusters complex wave forms with high fundamentals. Group11 seems to be based around the top end notes of the Indonesian flutes with the rest of the samples exhibiting the most complex wave forms of their groups.
Group12 is a very diverse cluster which appear to share little harmonic similarities. Group13 and
83 Group14 select only Rukai Bakararo 530HzG5 and 530HzG5 which are strongly consonant and dissonant respectively. viii
4.6 Brightness roughness features removed
1sec samples, four group clusters, British Pipe1Pipe2 and all overblown samples removed
Group 1
Rukai Bakararo 20% (6.7% of subset), Paiwan Gurare 40% (11.1% of subset), Thailand 40%
(20% of subset)
Group 2
Truku 10.5% (40% of subset), Koauau 7.9% (60% of subset), Indonesian 10.5% (50% of subset), Amis 2.6% (16.7% of subset), Philippine 15.8% (85.7% of subset), Thailand 13.1%
(50% of subset), Western 7.9% (60% of subset), Koauau TV 31.6% (100% of subset)
Group 3
Paiwan Gurare 12% (16.7% of subset), Indian 28% (100% of subset), Irish 28% (50% of subset), Amis 20% (83.3% of subset), Thailand 4% (10% of subset), Western 8% (40% of subset)
Group 4
Rukai Bakararo 20% (26.7% of subset), Rukai Gurare 30% (33.3% of subset), Truku 5% (10%
of subset), Koauau 10% (40% of subset), Indonesian 20% (50% of subset), Philippine 5%
(14.2% of subset), Thailand 10% (20% of subset)
Analysis
This test was designed as a supplemental experiment to further understand spectrum and centroid features Group1 clustered all high end frequencies with consonant intervals of the Paiwan Gurare. Group2 selected in an opposing cluster to Group1 by way of harmonics much closer to the first second and third harmonics, and appears to be strongly centred on the fundamental frequency. Group3 has clustered similar to group1 with much higher end harmonics, this group also included the most dissonant of Paiwan Gurare samples. Group4 is similar in clustering to group3. ix
84
Chapter 5: Conclusion
Summary of research results
Conclusion founded on the initial area of research concerning the organology of aboriginal musical instruments from Taiwan and New Zealand highlight two main factors, the first being the abundant quantity of Aerophones catalogued for each culture, this classification is further expanded by the vast assorted and variety of flutes and quasi-flute models. While the New Zealand Austronesian database doesn’t exhibit the cylindrical diversity found in its Taiwanese equivalent, which includes use of single and double pipe, drone and multi-holed combinations, it does show extensive use of deviation within a basic form. Playing techniques used in both cultures appear to have little in common, the nose flute techniques employed in divisions of the Taiwan dataset requires air pressure to the cylindrical resonant cavity with a considerable reduced pressure than that of the Maori equivalents, this is further disconnected by the Māori nose flute being highly disputed by leading researches in Maori ethnomusicology. Furthermore, the manufacturing, production techniques and materials employed in each set flutes samples appear to share no commonalities with the exception that the placement of finger holes which are governed by the manufactures individual digit ratio, this unit of measurement is found universally and can be no indication of relationships in production techniques. If there is a relationship to be found in these two sets of instruments it is to be found exclusively in the timbral identity produced by the Koauau of the Maori and the Headhunting flutes used predominately in the Northern Taiwan aboriginal cultures. This conclusion can be drawn confidently due to the analysis of sonic signatures exhibited by each sets spectrum, centroid, brightness and roughness features which recurred constantly in identical k-means clusters.
Another secondary finding in regards to the organology of Austronesian speaking Taiwan and New Zealand is the absolute lack of Membranophones used by both cultures (Table 3-I Indigenous Musical Instruments from Taiwan and New Zealand). Surprisingly, an instrument category that is generally considered to be an essential component of Polynesian material culture and ethnomusicology appears to be completely absent in both New Zealand and Taiwan. The
85 issue of the deficiency in drums, and percussion in general, is a possible subject of further enquiry for studies in Austronesian Ethnomusicology and could assist in the understanding of musical migration routes though the pacific.
Selected instruments from both cultures were both confirmed in the historical literature and in current aboriginal society. The acquisition of these Austronesian (and non-Austronesian) instruments, along with the musicians and instructors necessary demonstrate playing techniques was essentially a successful endeavour. Field work in both countries provided an acceptable database with which harmonic analysis could be performed as well as instruction on performance techniques with which further samples could be recorded. The acquisition of historical recordings which could have for validation, or dispute, of the modern harmonic signature was not achieved. While this was not essential to the principal concepts of this research, it would have strengthened the data by confirming the timbre produced by modern and historical recordings exhibited a similar timbral signature.
With regard to the principal concern of this research, the results clearly placed both Rukai and Paiwan double pipe flutes in the same cluster due to their unique sonic signature of emphasising 1f, 2f, 3f and for overblown dynamics samples, 4f of the fundamental. Likewise, when not vibrating in sympathy with the fundamental, the dissonant inharmonic intervals create an extremely complex waveform. It is possible the unique harmonic structure produced by double pipes influences the results in favour of extreme consonance and dissonance effect, though the continuity in placement of single pipe metallic instruments indicates towards accurate clustering procedures. With Truku, Maori and Philippine flutes constantly appearing in the similar clusters can be further understood as a relationship shared in the harmonic structure of this set of samples though this could not be absolutely assured without additional evidence to eliminate a clustering preference of simple harmonic vs. complex wave structure. While conclusions of the analysis and results can noticeably point to the assumption that Truku and Maori flutes clearly share a harmonic relationship unique with in this dataset, the inclusion of the Philippine flute within the same cluster may be seen as dubious, this is largely due to the lack of assurance in regards to the playing techniques employed for this instrument. Until such time as a musician familiar with this
86 instrument can confirm the playing techniques employed on this instrument, and therefore the timbre emitted, it would be prudent to question its inclusion within the database.
Contributions and limitations
In the course of investigating existing research pertaining to this study it became evident that no inquiry has dealt with the matter of harmonic analysis in regards to musical migration and cultural timbral preference. If this notion is indeed accurate it would signify this research as a preliminary investigation into an additional area of cultural musical comparisons, that would leave this research as being less of a contribution to Anthropology or Ethnomusicology and more of an investigation into the current state of musical digital processing and the abilities of music retrieval. From this stand point it could be said that this subject is ill-defined and still in it’s infancy as a conclusive science, that being the case, it would take substantial advancements in audio recognition in order to make decisive conclusions on timbral preference which would be absolutely necessary due to the subject of tonal identification being far too subjective for human identification and therefor impossible for any study to make a generalized conclusion on timbral groupings while being judged by an independent human source.
Several areas of concern were apparent immediately in the early analysis stages of this project, the first being the overall quality of the recorded audio samples, this was caused largely by the lack of equipment available and can only be rectified by assess to professional equipment and a standardised acoustic environment. The complications with this became noticeable when analysis
Several areas of concern were apparent immediately in the early analysis stages of this project, the first being the overall quality of the recorded audio samples, this was caused largely by the lack of equipment available and can only be rectified by assess to professional equipment and a standardised acoustic environment. The complications with this became noticeable when analysis