Professional Documents
Culture Documents
Conference Paper 40
Presented at the International Conference on Spatial and
Immersive Audio
2023 August 23–25, Huddersfield, UK
This conference paper was selected based on a submitted abstract and 750-word precis that have been peer reviewed by at
least two qualified anonymous reviewers. The complete manuscript was not peer reviewed. This conference paper has been
reproduced from the author’s advance manuscript without editing, corrections, or consideration by the Review Board. The
AES takes no responsibility for the contents. This paper is available in the AES E-Library (http://www.aes.org/e-lib), all rights
reserved. Reproduction of this paper, or any portion thereof, is not permitted without direct permission from the Journal of the
Audio Engineering Society.
Department of Mechanics and Vibroacoustics, AGH University of Science and Technology, Al. A. Mickiewicza 30, 30-059
Kraków, Poland
Correspondence should be addressed to Paweł Małecki (pawel.malecki@agh.edu.pl)
ABSTRACT
The paper outlines a process for producing and evaluating two mixes of the same musical composition, "Dancing
Ends" by Łukasz Pieprzyk, using spatial sound technologies: Dolby Atmos and Ambisonics. The composition was
recorded using overdub, multitrack techniques and the mixes were assessed through a direct evaluation method
based on average ratings on a 1 to 5 scale, in accordance with the ITU-R BS.128 standard. The criteria for evaluation
included selectivity, depth, width, and height of the sound stage, sound envelopment, brightness, and localization
quality and other aspects specific for the composition. These criteria were assessed in three different listening
systems and environments: 5.1, 7.1.4, and binaural. Listening tests revealed a preference for the Dolby Atmos mix
across all listening systems, even though the Ambisonic mix received higher ratings in several areas. The test group
comprised individuals with higher education in acoustics, basic sound production skills, and experience in listening
tests. The results were analyzed using the non-parametric Shapiro-Wilk test and t-Student or Mann-Whitney test.
AES International Conference on Spatial and Immersive Audio, Huddersfield, UK, 2023 August 23–25
Page 2 of 9
Malecki et al. Dolby Atmos vs Ambisonics
AES International Conference on Spatial and Immersive Audio, Huddersfield, UK, 2023 August 23–25
Page 3 of 9
Malecki et al. Dolby Atmos vs Ambisonics
AES International Conference on Spatial and Immersive Audio, Huddersfield, UK, 2023 August 23–25
Page 4 of 9
Malecki et al. Dolby Atmos vs Ambisonics
distinct evaluation methods: detection, ordinal, and samples were presented randomly. The first question
assignment procedures, each chosen according to the followed a single signal presentation principle (parame-
nature of the question. ter evaluation), whereas the subsequent questions used
a paired comparison (preference or difference evalua-
The detection method involved questions related to tion). Signal playback was sequenced consecutively for
compatibility evaluation (determining if samples are the first and last questions, while an option to switch
identical or different) or situations that necessitated a between signals was available for the rest.
choice (identifying the differing sample). The ordinal
evaluation was used for questions regarding ranking (in- Following the guidelines of the ITU-R BS.1284 for
tensity of a certain feature), preferences (better/worse), conducting subjective tests, the participant group com-
and similarity (most similar/different). The assignment prised a so-called ’expert group’ of at least ten individ-
method allowed for numerical estimation across differ- uals. For this experiment, twelve people were involved
ent types of scales. Listening tests were organized as for the 5.1 system and binaural listening, and ten peo-
surveys, where participants analyzed the material and ple for the 7.1.4 configuration. Each participant held a
made choices between the presented music excerpts. higher education degree in acoustics, had basic skills
Using Google Forms, a survey was composed of five in sound production, and previous experience in listen-
questions: ing tests. Some participants also held first or second-
degree music education. All were otologically normal,
meaning they were free from diagnosed diseases or
1. Rate the following sound properties:
pathologies of the auditory system.
(a) Selectivity To minimize the potential influence of the participants’
(b) Depth of soundstage emotions and attitudes on their judgment, questions
(c) Width of soundstage were precisely articulated, signals were equalized to
the same level (SPL A-weighted equal to 80 dB), and
(d) Height of soundstage all listeners received training prior to the listening tests.
(e) Sound immersion The initial preparation encompassed familiarization
(f) Clarity of sound with the survey structure, rules for presenting music
excerpts, and the methods of answering individual ques-
(g) Localization quality tions. Further clarification on the evaluation parameters
of the first question was provided in a document at the
2. Evaluate in which mix you can better locate the
commencement of the study, thus minimizing any mis-
flute?
understanding of the applied concepts.
3. Assess whether the piano’s position aligns pre-
cisely in both mixes? 4 Results
4. Evaluate in which mix you rate the balance be- Listening tests were conducted with the intent to com-
tween the string section and the brass section bet- pare the Ambisonic technique and Dolby Atmos in a
ter? selected speaker configurations. Based on the char-
acteristics of the constructed questions, a distinction
5. Which mix do you prefer?
was made between qualitative and quantitative vari-
ables. Depending on the features under examination,
Sound samples for each question were carefully se- data were classified on either an interval or nominal
lected to ensure signal diversity and reduce listener scale. The construction of the first question indicated
fatigue. Each music excerpt was designed to be di- quantitative variables assigned to the interval scale. Fol-
rectly related to a specific question. To maximize result lowing this assumption, it was necessary to investigate
accuracy by testing one person at a time, an interac- whether the results demonstrated characteristics of a
tive signal presentation was employed while the repeti- normal distribution (Shapiro-Wilk test), determine if
tion number was dependent on the listener’s decision. variables were correlated, and verify if the variances
All sound samples were logically arranged to avoid of variables across populations were equal (Levene’s
abrupt endings. For the preparation of testing material, test). Depending on the final assignment of data, an
AES International Conference on Spatial and Immersive Audio, Huddersfield, UK, 2023 August 23–25
Page 5 of 9
Malecki et al. Dolby Atmos vs Ambisonics
AES International Conference on Spatial and Immersive Audio, Huddersfield, UK, 2023 August 23–25
Page 6 of 9
Malecki et al. Dolby Atmos vs Ambisonics
AES International Conference on Spatial and Immersive Audio, Huddersfield, UK, 2023 August 23–25
Page 7 of 9
Malecki et al. Dolby Atmos vs Ambisonics
Fig. 8: Responds to the question Assess whether the Fig. 9: Responds to the question Evaluate in which mix
piano’s position aligns precisely in both mixes? you rate the balance between the string section
for 5.1, 7.1.4 systems and binaural listening and the brass section better? for 5.1, 7.1.4
(BIN) systems and binaural listening (BIN)
individual decisions of the sound engineer. It should analysis, and the interpretation of results obtained from
be noted that the ratings received in subjective tests the research. The subjective research stage involved
depended on many uncontrolled variables like time of conducting surveys of the participants about their psy-
the day etc. Another arbitrary, yet significant decision choacoustic impressions in two speaker configurations
that could impact the results was to start the mix with and binaural listening.
Dolby Atmos technology instead of Ambisonics, and
then attempt to replicate the achieved effect. Moreover, The subjective tests revealed a preference for the mix
the basic mix was made in the 5.1 system and was later made in Dolby Atmos technology in every listening sys-
checked and enhanced in a system dedicated to Atmos. tem, despite the fact that the Ambisonic mix was rated
higher in terms of seven criteria for evaluating the mu-
6 Summary sic material. In statistical terms, regarding technology,
in the 5.1 system, no significant differences between
The primary objective of this study was to conduct variables were demonstrated. For the 7.1.4 setup, sig-
a comparative analysis of spatial mixing executed in nificant differences were shown for the evaluation of
Ambisonic technology and Dolby Atmos. The music the scene width parameter, sound clarity, and local-
materials were subject to auditory evaluation, and the ization quality. In binaural listening, the scene width
scope of work included the description of mix real- parameter showed significant differences. The respon-
izations, the performance of listening tests, statistical dents stated that Dolby Atmos provides more precise
AES International Conference on Spatial and Immersive Audio, Huddersfield, UK, 2023 August 23–25
Page 8 of 9
Malecki et al. Dolby Atmos vs Ambisonics
[2] Kelly, J., Woszczyk, W., and King, R., “Are you
there?: A literature review of presence for immer-
sive music reproduction,” in Audio Engineering
Society Convention 149, Audio Engineering Soci-
ety, 2020.
[3] Zotter, F. and Frank, M., Ambisonics: A practical
3D audio theory for recording, studio production,
sound reinforcement, and virtual reality, Springer
Nature, 2019.
[4] Power, P. J., Future spatial audio: Subjective
evaluation of 3D surround systems, University of
Salford (United Kingdom), 2015.
[5] Rumsey, F., Griesinger, D., Holman, T.,
Sawaguchi, M., Steinke, G., Theile, G., and
Wakatuki, T., “Multichannel surround sound sys-
tems and operations,” The Audio Engineering So-
ciety, Tech. Rep. AESTD1001. 0.01-05, 2001.
[6] Francombe, J., Brookes, T., Mason, R., Wood-
cock, J., et al., “Evaluation of spatial audio re-
production methods (part 2): analysis of listener
preference,” Journal of the Audio Engineering
Society, 65(3), pp. 212–225, 2017.
Fig. 10: Responds to the question Which mix do you [7] Malecki, P., Piotrowska, M., Sochaczewska, K.,
prefer? for 5.1, 7.1.4 systems and binaural and Piotrowski, S., “Electronic music production
listening (BIN) in ambisonics-case study,” Journal of the Audio
Engineering Society, 68(1/2), pp. 87–94, 2020.
[8] “IEM Plug-in Suite,” https://plugins.
source localization. However, they did not agree on the iem.at/, 2023.
aspect of localization accuracy in both mixes.
Another very important conclusion is that the ambigu- [9] “Kotłownia Studio,” https://kotlownia.
ity of the results obtained suggests that the difference agh.edu.pl/, 2023.
between the systems is not significant. This opens up [10] “Dancing Ends for Symphony Orches-
possibilities for the production of high-quality spatial tra and Piano by Łukasz Pieprzyk,”
signals using open technology and free tools. https://open.spotify.com/track/
Currently, spatial sound is undergoing significant tech- 3fCCmIQvxLeAb0WdIuSgtj, 2023.
nological development and is being used more com-
monly. The topic addressed in this work requires a [11] “120dB ATMOS,” https://www.120db.
broader perspective in terms of different music genres pl/atmos/, 2023.
and listening systems. The conducted research could
serve as a basis for further analysis.
References
[1] “About Spatial Audio with Dolby Atmos in
Apple Music,” https://support.apple.
com/en-us/HT212182, 2023.
AES International Conference on Spatial and Immersive Audio, Huddersfield, UK, 2023 August 23–25
Page 9 of 9