Sound documentation of historical pipe organs: a case study Christophe d’Alessandro LIMSI-CNRS - University Paris XI Orsay, France with Brian F.G. Katz, Guillaume Nief & Michel Barat
1
MAN-GSAM’06
Contents ¾ The aims of sound documentation: conservation, “organologie”, musicology and music. ¾Multilayered documentation: room acoustics, organ sampling, portraits and styles. ¾First conclusions and projects
2
MAN-GSAM’06
The aims of sound documentation ¾Conservation: ¾Preservation of cultural heritage ¾Sound archives ¾Documentation for restorations ¾ period organ design
¾Musicology: ¾Stylistic variations and features ¾Historic and national styles, evolution, diffusion ¾Instruments and repertoire, performance history
¾Music: ¾Virtual historical organs for training and education at home and music schools ¾Concert rehearsal on specific organ avatars 3
MAN-GSAM’06
A paradox in historical instruments documentation
¾ archives documentation ¾ technical documentation ¾ copy of instruments and building techniques ¾ Copy of ancient material ¾ … but almost no sound documentation (musical recordings excepted)
4
MAN-GSAM’06
¾ … but almost no sound documentation (musical recordings excepted) and obviously musical recordings are not enough, because: ¾They are sound processed for pleasantness ¾They are performances and not measurements ¾Recording techniques are not documented ¾They are only partial: only a part of the instrument is actually played 5
MAN-GSAM’06
Documentation projects ¾Records of Listed historical instruments: ¾ Commission Supérieure des Monuments Historiques: a few (significant) instruments
¾Research projects:
¾Tonal portraits, loudness analyses ¾Plenum comparison, stop comparisons ¾Transient analyses, steady state analyses
¾Virtual organ projects and products:
¾Hauptwerk ™ , MyOrgan ©, Miditzer™ other virtual organ projects ¾Plenty of sampled historical organs, various recording conditions and qualities
6
MAN-GSAM’06
Organ landscape in France ¾ Commission Supérieure des Monuments Historiques (MH) ¾ ~858 listed MH instruments ¾ ~606 listed MH cases ¾ ~ 8000 organs in public buildings (partly 14th-21th century) ¾ ~5500 cinemas,
7
MAN-GSAM’06
Organ landscape in France ¾ Commission Supérieure des Monuments Historiques (MH) ¾ ~858 listed MH instruments ¾ ~606 listed MH cases ¾ ~ 8000 organs in public buildings (mostly churches or temples, instruments partly 14th-21th century) ¾ ~5300 cinemas, ¾ 60.6M (1 organ / 7500 ) 8
MAN-GSAM’06
Organ landscape in France XIVth – XVth century
9
MAN-GSAM’06
Organ landscape in France XVIth – XVIIth century
10
MAN-GSAM’06
Organ landscape in France XVIIIth – XIXth century
11
MAN-GSAM’06
Organ landscape in France XXth – XXIst century
12
MAN-GSAM’06
Organ landscape in France English organs in France
13
MAN-GSAM’06
Ste Elisabeth : Multilayered sound documentation Room acoustics Layer:
o Impulse, noise, organ response recordings o Perceptive parameters analysis
Organ sampling layer:
o individual pipe recordings o multi-microphones / multi-positions
Tonal features Layer:
o pipes and stops tonal features and portraits o transient analyses, spectral analyses
14
MAN-GSAM’06
The Organ • Built by Antoine Suret 1853 • Listed MH and restored 1994-99 • • • •
39 stops 42 registers 3 keyboards (54 notes) 1 pedal-board (30 notes)
• 16 reed stops – – – – – –
Two 16’ bombard Five 8’ trumpets Three 4’ clairons 2 hautbois Clarinette, cor anglais voix humaine, basson
• In total, over 2500 pipes
15
MAN-GSAM’06
Room acoustics and the organ ¾The Organ is unique primarily due to its size as compared to the room, and other traditional musical instruments. ¾Quantitative measures of the effect of room acoustics on the subjective impression of the organ
16
MAN-GSAM’06
The Church : L’Eglise St. Elisabeth du Temple, Paris • • • • •
Constructed mid XVIIth Century Enlarged mid XIXth Century Single glazed windows at high level Floor of wood and stone Stone walls Acoustic envelope
• • • •
Central general seating for ~300 Floor area ~1200m2 Main central volume ~6300m3 Central axis ceiling height of 17m • Peripheral ceiling height of 5.5m • 17 x 26 x 45 m
Organ
17
MAN-GSAM’06
Room acoustics and the organ • Short prelude and fugue in F, BWV 556, J.S. Bach. • Registration : positif de dos: flûte 8', prestant 4', bourdon 8', Kéraulophone 8', nasart 2'2/3
• Inside the case • Last row • First raw • In front of altar
18
MAN-GSAM’06
Measurements : Source / Receiver Positions • Source 1 : Point source – Balloon impulse source : Floor level
• Source 2 : Point source – Loudspeaker source : Organ level
• Source 3
• 11 receiver positions at 3m intervals starting 3m from the rear-wall / organ façade. p3
p11
– Entire organ excited using all stops, coupled keyboards, and all keys/pedals
Loudspeaker, located atop the positive chest Balloon source located at the same position as the loudspeaker in plan 19
17m ceiling MAN-GSAM’06
The Equipment • Acoustic measurements were made using a KEMAR dummy head and torso. • Binaural recordings are obtained through microphones located at the entrance to the ear canal. • The head is always aligned in the same direction, regarding the same point at the center of the room. • Data acquisition was direct to disc using a laptop and external A/D converters operating at 44.1kHz and 24bit.
20
MAN-GSAM’06
Measurements : Reverberation Time 4.0
• Room Characterization • Reverberation time, T20, was calculated at all measurement positions • Average of 2 repetitions and 2 ears • Strong mid-freq reverberation
Reverberation Time- T20 (sec)
3.5
3.0
2.5
2.0 p01 p02 p03 p04 p05 p06 p07 p08 p09 p10 p11
1.5
1.0
0.5
0.0 125
• Low-frequency dropoff • Minimal variations between positions 21
250
500
1000
2000
4000
Octave band center freq (Hz)
125
250
500
1000
2000
4000
mean(T20)
2.70
3.10
3.68
3.68
3.03
2.14
std(T20)
0.18
0.12
0.10
0.07
0.04
0.03
Spatial average and corresponding standard deviation of reverberation time, T20, in octave bands.
MAN-GSAM’06
Measurements : IACC (Description) • Inter-aural Cross Correlation function
⎡ ⎢ IACC t = max ⎢ IACC (τ ) = ⎢ ⎣
⎤ ∫t1 [ p L (t ) ⋅ p R (t + τ )]dt ⎥ ⎥ t2 t2 2 2 ( ) p t dt p ∫t1 L ∫t1 R (t )dt ⎥⎦ (− 1 < τ < 1) t2
• • • •
Measure of the similarity between signals at the left and right ear Ideally, a frontal source in an anechoic chamber gives IACC = 1 The less similar (smaller IACC) the more “spacious” Octave band Frequency averaging and different integration times are found to correlate with different types of subjective impressions • Average of octave band IACC values for 500,1000, 2000 Hz : IACC3
• IACCA3: ALL of the response 0 → ∞ • IACCE3: EARLY part : 0 → 80 msec • IACCL3: LATE part : 80 → (1000 msec,∞) 22
Spatial impression Apparent Source Width Listener Envelopment MAN-GSAM’06
Measurements : IACC (Organ Source)
23
0.8
Speaker IACC_A3 0.7
Speaker IACC_E3 Organ IACC_A3
0.6
Organ IACC_E3
0.5
IACC
• Typical values for good concert halls : IACCE3 = 0.6 0.7 • IACCA3 and IACCE3 measurements for the loudspeaker and full organ sources. • IACCA3 shows very similar values (relating to Listener Envelopment) • IACCE3 shows a reduction for the organ source, indicating a larger perceived source width for the majority of measurement positions.
0.4 0.3 0.2 0.1 0.0 p01
p02
p03
p04
p05
p06
p07
p08
p09
p10
p11
Position
MAN-GSAM’06
The “Positif de dos” • 8 stops: ¾ Flûte 8’ (stopped 1rst octave, part in facade) 54 pipes ¾ Kéraulophone 8’ 42 pipes ¾ Prestant 4’ (part in facade) ¾ Nasard 2’ 2/3 (part in facade) ¾ Bourdon 8’ ¾ Trompette 8’ ¾ Basson-Hautbois 8’ ¾ Clairon 4’
420 pipes, 162 reed pipes and 258 flue pipes
24
MAN-GSAM’06
Organ sampling methodology
¾ 2 positions:
¾ Inside: in the center of the case ¾ Outside: about 2 meters in front of the case, 3 meters on Ut# side of the tribune
¾ All the pipes for all stops ¾ Calibrated recordings ¾ Full duration tone, until complete damping 25
MAN-GSAM’06
Organ sampling methodology
¾ Memory : ¾ Mono : 44.1 KHz X 6s = 300k sample /pipe (1.2 Mbyte / 32 bit). ¾ 60 Mb / stop ¾ 2 positions ~ 1 Gb 26
MAN-GSAM’06
Sampling : inside
¾ Near acoustic field ¾ Tremendous variability depending on the recording position ¾ Good rendering of initial transient ¾ Weak influence of room acoustics ¾ The “montre” pipes (façade) are … outside ¾ The voicer’s point of hearing 27
MAN-GSAM’06
Sampling : outside
¾ ¾ ¾ ¾ ¾ 28
Far acoustic field Spatial effects Poor rendering of initial transient Strong influence of room acoustics The audience’s point of hearing MAN-GSAM’06
Sampling: evaluation ¾ Real organ: ¾ Inside the case ¾ A few meters in front of the case
¾ Virtual organ: ¾ Inside the case ¾ A few meters in front of the case ¾ Inside the case + reverb
29
MAN-GSAM’06
Sampling: limitations ¾ Interaction between pipes ¾ ¾ ¾ ¾
Synchronization Non linear acoustic mixtures Pipe radiation patterns Wind variation when many pipes are played
¾ Noise addition ¾ Blower ¾ Action
¾ Point of hearing
¾ Radiation pattern: wide acoustic source (10mX10m) ¾ Room acoustics
30
MAN-GSAM’06
Tonal features ¾ SPL ¾ Oscillograms and Spectrograms ¾ Transient building ¾ Dominant harmonics ¾ Energy of the first harmonics ¾ Energy in critical bands ¾ Spectral centre of gravity 31
MAN-GSAM’06
Oscillograms
32
MAN-GSAM’06
Spectrograms (0-22kHz)
33
MAN-GSAM’06
Spectrograms (0-8kHz)
34
MAN-GSAM’06
SPL profiles (out)
35
MAN-GSAM’06
SPL profiles (in)
36
MAN-GSAM’06
Spectral centre of gravity (out)
37
MAN-GSAM’06
Spectral centre of gravity (in)
38
MAN-GSAM’06
Harmonic amplitudes (out)
39
MAN-GSAM’06
Dominant harmonics (out)
40
MAN-GSAM’06
Dominant harmonics (in)
41
MAN-GSAM’06
Critical band energy (out)
42
MAN-GSAM’06
Starting transient (in) • Global time of establishment of steady state • Amplitude of harmonics evolution during starting transient
43
MAN-GSAM’06
Tonal features • Relative SPL profiles ÅÆ balance between stops • Individual SPL profiles ÅÆ low pitch / high pitchbalance • Dominant harmonic H1-H2 ÅÆ pipe scales (narrow/medium/large) • Spectral richness (CGS…) ÅÆpipe scales – reeds/flue pipes • Formant structure ÅÆ reed scale / resonator shape • Missing H2 ÅÆstopped pipes • Critical band spectra ÅÆ brightness • Spectrograms ÅÆ “registers” within a same stop 44
MAN-GSAM’06
Tonal portrait: example of the flûte 8’
• medium pipe scaling • Stopped first octave • 3/4 “registers” • Stronger for higher notes • Globally constant SCG • Note to note variability
45
MAN-GSAM’06
Conclusions ¾ Signal processing approach to acoustic documentation ¾ 3 layered documentation: room acoustics, sampling, tonal features features ¾ Most tonal features seems globally robust to recording position (although large individual variations are between individual pipes) ¾ Input of organ builders, organists and experts is needed for refining the documentation methodology, especially for perceptual analyses 46
MAN-GSAM’06
Projects ¾ Multipoint recordings all the pipes + typical registrations ¾ Sampling assessment ¾ Use perceptual measures (loudness instead of SPL, etc.) ¾ Room acoustics measurement assessment ¾Recording with room acoustic attenuation (large curtain) 47
MAN-GSAM’06