Auralisation-based research needs anechoic recordings. Where opera is concerned, anechoic material is limited to a few recordings. Using the approaches of previous recordings, the present work shows methods and results of a recording set focused on the Italian melodrama period. The chosen motifs come from scores of Donizetti, Verdi, and Puccini in order to consider the most performed composers and the peculiarities of their styles, which span from the early 19th century to the 20th century. Excerpts from three operas have been played by professional musicians and soloists with an array of twelve microphones in a damped room, which can be considered anechoic. The multi-track files of the recorded materials are freely available for academic purposes.

Anechoic recordings were used since the 1970s for listening tests based on auralisations.1 Nowadays, several research fields need anechoic recordings for: preference ratings, obviously,2,3 but also heritage preservation,4,5 acoustical design,6 phenomenological issues,7 and more.

Earlier anechoic recordings were recorded in a BBC anechoic chamber8 consisting of compositions for small orchestras. Despite its low dynamics, these recordings were widely used in early subjective investigations on room criteria.1 In 1988, a large orchestra was recorded in a damped concert hall.9 The study was mainly oriented towards the comparison of different miking techniques and the subjective evaluation of convolution techniques. In the years 1987–1992, the Archimedes project recorded excerpts for soloists (speakers and musicians) in various acoustic conditions (anechoic chamber, listening room, studio, church).10 The goal of the project was to quantify the “subjective influence of listening room acoustics and loudspeaker directivity on the reproduced sound.” In the ERATO project, which investigated the acoustics of the ancient theatres, ancient Greek singing pieces were recorded11 by scholars of Ferrara University. The same research group was involved in the CHARISMA project, which was focused on ancient Islamic and Byzantine intangible heritage.12 The small ensemble or soloist was recorded for the first time using an array of five omnidirectional microphones.

In almost all of the previous works, musicians were recorded while playing together, as in an ensemble. On the contrary, Vigéant et al.13 used multichannel techniques to record musicians playing solo, following a video recording of the conductor as sync source. The musicians were recorded individually in the anechoic room with the same cubic array used by ERATO and CHARISMA projects. Two or three tracks were recorded for each string part, with one track for the remaining parts. Some years later, Pätynen et al.14 recorded symphonic and opera music by using an approach similar to the previous study but extended the number of musicians and the miking spatial resolution: the orchestra parts and a singer were recorded individually in the anechoic room of Aalto University using an array of twenty-two microphones. Anechoic pieces for choir were also recorded by Freiheit;15 the pieces spanned several centuries of choral music production, but the voices were recorded simultaneously as an ensemble.

The most recent years' trends follow the Vigéant-Pätynen multi-track workflow: Pedrero et al. recorded coral pieces from Mozarabic tradition;16 Woirgardt et al. recorded Flamenco, Blues, and Pop pieces for guitar and voice;17 Böhm et al. recorded Beethoven's 8th Symphony;18 Théry and Katz recorded string quartets, jazz “manouche,” and standard jazz pieces for the little ensemble.19 It should be noted that recently the recording quality was increased in order to achieve a pleasurable listening experience, high dynamics of the signal, and low noise levels. All of these aspects are critical to obtain realistic auralisations. For these reasons, Patýnen's recordings were denoised and re-distributed in 2016.20 

A partial taxonomy of anechoic recordings provided by previous projects is summarised in Table 1. Table 2 lists the musical terms in Italian and Table 3 lists the anechoic recordings of the present work.

The present work aims to extend the availability of recorded anechoic material by introducing three multitrack recordings of voice parts of an Italian opera. A dodecahedral array of microphones with a 1.1 m radius were set in a listening room damped with porous material. The excerpts were recorded by musicians and soloists one track by one track under the supervision of a conductor, who listened and re-recorded each track until a proper orchestral sound was reached. The noises not produced from instruments, such as hisses, breathes, and body movements, were manually erased in the post-processing phase. All the tracks of the array were exported in .wav format and uploaded to a free repository.

In order to select representative opera material to be recorded, a preliminary study was conducted analysing the statistics of opera performances in the 2015–2016 season in the most important opera houses.21 Taking into account these statistics, the search domain of music material spans from the early 19th century to mid 20th century, in which the opera was progressively “standardising.” Before this period, the baroque opera included recitativo parts written for cembalo only and less instruments. After this period, modern and contemporary compositions are protected by copyright. The music material for the present study was chosen to represent the most performed Italian composers and their styles: Donizetti for the early 19th-century style, Verdi as the “classic form” of 19th-century opera, Puccini for the 20th-century evolution. In the following each excerpt is briefly described.

The first excerpt is an aria (“Come Paride vezzoso”) from the opera L'elisir d'amore by Gaetano Donizetti (1797–1848). The excerpt represents the belcanto in the Italian Opera: a cavatina for coloritura baritone. In the final cadenza and some agility passages which are peculiarities of the Italian opera style. In this motif there are parts of other soloists (a tenor and two sopranos) and a choir. The score shows several tempo variations: the “larghetto” in 3/4 then the “andantino” in 4/4, the free “colla voce,” which resolves to the initial time (“a tempo”) in the final (see Fig. 1).

The second recording is extracted from the opera Il trovatore by Giuseppe Verdi (1813–1901). In the cabaletta “Di tale amor, che dirsi” the soloist figure is a lyric soprano in whom voice weight and agility coexist. In this excerpt the Verdian orchestra is complete: strings, full woodwinds with ottavino, full brasses with bass trombone (cymbasso). The final part presents a crescendo with a wide dynamic range and an accelerated tempo.

In order to complete the temporal evolution of the Italian Opera, the third motif is the romanza “Oh Mio Babbino Caro” from the opera Gianni Schicchi by Giacomo Puccini (1858–1924). Here the soprano sings with a slight voice and the music is a siciliana played by a large string section, a few woodwinds, and a harp. The recording also includes the initial tone clusters for strings and brasses, usually played in the recital version of the romanza.

The recording room was the listening room of the University of Bologna (dimensions: 7.5 × 5 × 3 m) that was damped with high density fiberform (80 kg/m3). Porous material covered all the surfaces, including the floor: performers played above a metallic grid. The porous materials and box-in-a-box structure made by gypsum boards allowed a negligible reverberation in a wide range of frequency—less than 50 ms at 125 Hz. The radius of dodecahedral microphone array measured 1.1 m. This distance is lower the critical radius of reverberation, allowing more than 10 dB of direct-to-reverberant ratio.

Professional musicians from a young orchestra and soloists were asked to join this research. Some musicians had previous experience in recording contemporary and pop music too. A total of twelve musicians and three soloists were involved in the recordings.

The arrangement of the recording follows the Vigeant-Pätynen framework: one musician at a time, for each instrument, played all the score parts, following a reference video of the conductor with a pianist; during the recording takes, the musicians heard the sync piano and the previous tracks through closed headphones.13,14 The conductor attended all the recording sessions in order to evaluate the effectiveness of the single takes on the general impression of the orchestra. Recordings were monitored from a control room, with a talk-back connection with the conductor. Double basses and cellos were recorded in session no.1, followed by second violins and violas. Only a single track of the first violin was recorded as reference for the intonation. One violinist played all the tracks of the first and second violin parts. In the second session, brasses were recorded: horns, trombones, and trumpets. From session No. 3 onward—which involved the woodwinds—the piano track has been removed from the audio monitoring heard by musicians during recordings. In the fourth session, all parts of the first violins were recorded with the right orchestral timbre and dynamics. In the last session, the harp was recorded and some takes were overdubbed. A comparison between the orchestral part of the previous recordings and the present ones is shown in Table 4.

Each musician was placed at the center of the microphone array (see Fig. 2) and played and repeated his/her part until earning the approval of the director. The tracks were recorded without stops and overdubs. The recordings were monitored in real time from both conductor and musician. The conductor was placed in a control room, listening to the real-time performance mixed with the previous tracks through near field monitor or closed headphones. Musicians could listen to themselves and to some or all of the previous recording tracks, depending on their requirements. Following musician requirements, the right channel only was powered to the headphones for self-hearing. Headphone listening was individually set for each performer in order to allow an adequate self-monitoring and—at the same time—without compromising the recordings with the sound coming from the headphones. After each take musicians could speak with the conductor through talk-back connection in order to improve the execution. Each take was generally recorded three or four times until the conductor approved.

With respect to the work of Pätynen et al., where each string part has been recorded once,22 in the present work the strings have been recorded several times (see Table 4). The same violinist played the tracks of the first and the second violin, performing all the different sessions. The same was done with brasses and other multi-track parts. Any digital-processing was added to increase the “chorus” effect, e.g., through phase shifting. It should be noticed that the conductor did not use a metronome during recordings, not even in the reference recordings of piano and voice, in order to keep the feeling of a “real” execution.

Audio-Technica AT4050 microphones have been chosen thanks to the good recording capability and low noise characteristics. The microphones have been used in omnidirectional configuration and have been pre–amplified and A/D converted 48 kHz 24 bit through a RME Micstasy, set with about 35 dB of internal gain. A pad attenuation of −15 dB was used for the trombone takes only.

The tracks were recorded with reaper software,23 and then post-processed in the same environment. The tracks were manually checked to delete breaths and hisses before the sound attacks and after the releases. Breaths during the notes were not erased, in order to preserve the instrumental timbre. Background noises during silence—due to monitoring through headphones, body movements, etc.—were manually deleted. The recorded tracks were not equalised. The tracks were exported in 12-track .wav format, at 48 kHz/24 bit, the number of each sub-track corresponding to a microphone of the dodecahedral array. Multichannel files and recording details are available for free.24 

Anechoic recordings of opera excerpts were presented. Soloist singers and ensemble musicians were recorded following procedures similar to those found in the existing literature. Multitrack recordings were made using a dodecahedral array of large-diaphragm microphones. The instrumental parts were performed in different sessions, recording one part at a time and then overdubbed, in order to reach a proper orchestral sound. The recorded material is free for research purposes.

The author gratefully acknowledges all the people involved in this project: the conductor Jacopo Rivani and the soloists and musicians of the Corelli ensemble.

1.
M. R.
Schroeder
,
D.
Gottlob
, and
K. F.
Siebrasse
, “
Comparative study of European concert halls: Correlation of subjective preference with geometric and acoustic parameters
,”
J. Acoust. Soc. Am.
56
,
1195
1201
(
1974
).
2.
A.
Kuusinen
, “
An anechoic audio corpus for room acoustics and related studies
,” in
Proceedings of the EAA Joint Symposium on Auralisation and Ambisonics
, Berlin, Germany (3–5 April 2014).
3.
M. S.
Lawless
and
M. C.
Vigeant
, “
Effects of test method and participant musical training on preference ratings of stimuli with different reverberation times
,”
J. Acoust. Soc. Am.
142
,
2258
2272
(
2017
).
4.
S.
Weinzierl
and
S.
Lepa
, “
On the epistemic potential of virtual realities for the historical sciences. A methodological framework
,” in
Augmented Reality. Reflections on Its Contribution to Knowledge Formation (Berlin Studies in Knowledge Research 11)
, edited by
J. M.
Ariso
(
De Gruyter
,
Berlin
,
2017
), pp.
61
80
.
5.
D.
D'Orazio
,
S.
De Cesaris
,
F.
Morandi
, and
M.
Garai
, “
The aesthetics of the Bayreuth Festspielhaus explained by means of acoustic measurements and simulations
,”
J. Cult. Herit.
34
,
151
158
(
2018
).
6.
D.
Thery
,
V.
Boccara
, and
B. F.
Katz
, “
Auralization uses in acoustical design: A survey study
,”
J. Acoust. Soc. Am.
145
(
6
),
3446
3456
(
2019
).
7.
D.
D'Orazio
and
M.
Garai
, “
The autocorrelation-based analysis as a tool of sound perception in a reverberant field
,”
Riv. Estet.
66
,
133
147
(
2017
).
8.
A. N.
Burd
, “
Non-reverberant music for acoustic studies
,”
The British Broadcasting Corporation Report No. 1969/17
(
1969
).
9.
T.
Hidaka
,
K.
Kageyama
, and
S.
Masuda
, “
Recording of anechoic orchestral music and measurement of its physical characteristics based on the auto-correlation function
,”
Acustica
67
,
68
70
(
1988
).
10.
V.
Hansen
and
G.
Munch
, “
Making recordings for simulation tests in the Archimedes project
,”
J. Audio Eng. Soc.
39
(
10
),
768
774
(
1991
).
11.
Z.
Yuksel
and
S.
Erdogan
, “
Virtual conservation of acoustical heritage: CHARISMA and ERATO Projects
,” in
Proceedings of Forum Acusticum
, Budapest (
2005
).
12.
Z.
Karabiber
, “
A new approach to an ancient subject: CHARISMA Project
,” in
Seventh International Congress on Sound and Vibration
, Garmisch-Parterkirchen (
2000
).
13.
M. C.
Vigéant
,
L. M.
Wang
,
J. H.
Rindel
,
C. L.
Christensen
, and
A. C.
Gade
, “
Multi-channel orchestral anechoic recordings for auralizations
,” in
Proceedings of the International Symposium on Room Acoustics
, Melbourne (
2010
).
14.
J.
Pätynen
,
V.
Pulkki
, and
T.
Lokki
, “
Anechoic recording system for symphony orchestra
,”
Acust. Acta Acust.
94
(
6
),
856
865
(
2008
).
15.
R.
Freiheit
, “
Creating an anechoic choral recording
,” in
Proceedings of the International Symposium on Room Acoustics
, Melbourne (
2010
).
16.
A.
Pedrero
,
M.
Pollow
,
P.
Dietrich
,
G. K.
Behler
,
M.
Vorlaender
,
C.
Díaz
, and
A.
Díaz
, “
Mozarabic chant anechoic recordings for auralization purposes
,” in
Proceedings of the VIII Congresso Ibero-americano de Acustica
, Evora, Portugal (1–3 October
2012
), Vol.
80
, p.
8S
.
17.
M.
Woirgardt
,
P.
Stade
,
J.
Amankwor
,
B.
Bernschütz
, and
J.
Arend
, “
Cologne University of Applied Sciences—Anechoic Recordings
,” (2013), www.audiogroup.web.fh-koeln.de (Last viewed January 31,
2020
).
18.
C.
Böhm
,
D.
Ackermann
, and
S.
Weinzierl
, “
Eine mehrkanalige und nachhallfreie Aufnahme von Beethovens 8. Sinfonie
,” in
Fortschritte der Akustik—44. Jahrestagung für Akustik (DAGA)
, München (
2018
).
19.
D.
Théry
and
B.
Katz
, “
Anechoic audio and 3D-video content database of small ensemble performances for virtual concerts
,” in
International Congress on Acoustics (ICA)
, September 2019, Aachen, Germany.
20.
M.
Miron
,
J.
Carabias-Orti
,
J. J.
Bosch
,
E.
Gómez
, and
J.
Janer
, “
Score-informed source separation for multi-channel orchestral recordings
,”
J. Elec. Comput. Eng.
2016
,
836350
.
21.
D.
D'Orazio
,
S. De
Cesaris
, and
M.
Garai
, “
Recordings of Italian opera orchestra and soloists in a silent room
,”
Proc. Mtgs. Acoust.
28
,
015014
(
2016
).
22.
T.
Lokki
, “
How many point sources is needed to represent strings in auralization?
,” in
International Symposium on Room Acoustics
, Seville, September 10–12,
2007
.
23.
Reaper v. 5.20, http://www.reaper.fm (Last viewed January 31,
2020
).
24.
D.
D'Orazio
,
Anechoic Recordings of Italian Opera
.