Conference Paper
Subjective Evaluation of MP3 Compression for Different Musical Genres
Conference: Proceedings of the 127th Convention of the Audio Engineering Society (AES)
Figures in this publication
| Get notified about updates to this publication Follow publication |
Full-text
Available from: Amandine Pras
Audio Engineering Society
Convention Paper
Presented at the 127th Convention
2009 October 9–12 New York, NY, USA
The papers at this Convention have been selected on the basis of a submitted abstract and extended precis that have been peer
reviewed by at least two qualified anonymous reviewers. This convention 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.
Additional papers may be obtained by sending request and remittance to Audio Engineering Society, 60 East 42
nd
Street, New
York, New York 10165-2520, USA; also see www.aes.org. 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.
Subjective evaluation of mp3 compression
for different musical genres
Amandine Pras
1
, Rachel Zimmerman
2
, Daniel Levitin
2
and Catherine Guastavino
1
1
Centre for Interdisciplinary Research in Music Media and Technology,
Multimodal Interaction Laboratory, McGill University, Montréal, Québec, H3A 1EA, Canada
amandine.pras@mcgill.ca, catherine.guastavino@mcgill.ca
2
Centre for Interdisciplinary Research in Music Media and Technology,
Department of Psychology, McGill University, Montréal, Québec, H3A 1B1, Canada
rachel.zimmerman@mail.mcgill.ca, daniel.levitin@mcgill.ca
ABSTRACT
Mp3 compression is commonly used to reduce the size of digital music files but introduces a number of potentially
audible artifacts, especially at low bitrates. We investigated whether listeners prefer CD quality to mp3 files at
various bitrates (96 kb/s to 320 kb/s), and whether this preference is affected by musical genre. Thirteen trained
listeners completed an A/B comparison task judging CD quality and compressed files. Listeners significantly
preferred CD quality to mp3 files up to 192 kb/s for all musical genres. In addition, we observed a significant effect
of expertise (sound engineers vs. musicians) and musical genres (electric v.s acoustic music).
Pras et al.
Subjective evaluation of mp3 compression
AES 127th Convention, New York, NY, USA, 2009 October 9–12
Page 2 of 7
1. INTRODUCTION
The Motion Picture Experts Group (MPEG) designed
the coder-decoder ("codec") soundfile format mp3 in
1991 as a lossy audio standard to reduce the size of
digital music files. In 1991, few people had Internet
connections and those who did used dial-up services
with typical transmission rates of 5kB/s. Recordable
compact discs were not even available in many high-end
recording studios. For these reasons, reducing the file
size was especially important at that time. Whereas mp3
compression reduces the amount of storage space, the
trade-off is a potential decrease in the sonic quality of
the audio files. Although these restrictions have eased,
many people listen to all their music in compressed
formats. In addition, e-tailers encourage people to buy
compressed music over the Internet, while the mastering
of music productions is still performed at CD quality or
even higher quality.
When encoding in mp3, users choose a bitrate
(measured in kbits/s) that determines the compression
ratio and subsequently the size of the encoded file.
During the encoding process, the audio content is first
decomposed into 32 frequency sub-bands that are
processed based on perceptual models [6]. These
models are based on psychoacoustic mechanisms such
as the frequency response of human hearing, as well as
frequency masking and temporal masking. Then, each
sub-band is encoded with a bit allocation determined by
the number of available bits (bitrate) and the audio
content of the sub-band [1]. This technique theoretically
introduces errors in high frequency resolution, as well
as noise due to the quantification errors. Regarding
audible artifacts introduced by perceptual coders,
Brandenburg [1] states that “…the signal may be
sounding distorted, but not like harmonic distortion,
noisy […] and rough”.
Previous research has compared different types of audio
compression algorithms including mp3 with one another
[7]. However, perceptual differences between CD
quality and mp3 compression have received limited
attention. Ruzanski [3] observed a significant
interaction between the genre of the clip and how
tolerant it was to compression before listener preference
would be influenced, for bitrates ranging between 32
and 192 kb/s. Furthermore, Sutherland [8] observed that
expert listeners, defined as professional sound engineers
with more than ten years of experience, significantly
preferred CD quality to compressed files even at very
high bitrates (up to 320 kb/s), whereas Salimpoor [5]
found that average listeners were only able to
distinguish between very low bitrates (96 kb/s) from CD
quality.
Based on these studies, we hypothesize that trained
listeners who have studio experience (musicians and
young sound engineers with less than ten years of
experience) will be able to discriminate between CD
quality files and mp3 compressed files and that they will
tend more to prefer CD quality files. Furthermore, we
seek to know if the results are a function of musical
genre and/or listeners’ expertise. Finally, we wish to
investigate if trained listeners can identify perceptual
sound artifacts that are introduced by mp3 compression
and if these artifacts depend on the musical genre.
2. METHODS
2.1. Participants
Thirteen trained listeners, eight males and five females,
took part in the study and received $20 for their
participation. All subjects passed a standard audiometric
for normal hearing. The participants’ mean age was 28
(S.D. = 5.6), and mean length of studio experience was
six years (S.D. 5.2). They had different sources of
studio listening expertise: four as musicians, eight as
sound engineers (three young professionals and five
students in a sound recording masters degree program),
and one as a researcher
1
. All participants reported
having taken music lessons for 15 years on average
(S.D. 4.7) except for one professional sound engineer,
who hadn't received musical training. The participants
reported listening to music for an average three hours a
day (S.D. 2.8) at home or work, and all musical genres
were represented. They reported listening to music in
lossy compressed format (mp3 or others) 36% of the
time on average (S.D. 23.6) as opposed to 64% of the
time for uncompressed or lossless compression formats.
2.2. Musical excerpts
We selected five short musical excerpts (one musical
phrase ranging from 5 to 11 sec.) in CD quality (44.1
KHz, 16 bit) representative of different musical genres:
pop, metal rock, contemporary, orchestra and opera (see
detail of the musical excerpts in Table 1).
1
Two of the co-authors were participants in the study, one
professional sound engineer (AP) and one research scientist (CG).
Pras et al.
Subjective evaluation of mp3 compression
AES 127th Convention, New York, NY, USA, 2009 October 9–12
Page 3 of 7
Musical
genre
Name of the
piece (tune)
Composer
Performers
(band/orchestra)
Additional Information
Pop
Irish Green
Bart Moore
Slings & Arrows
Produced by Daniel Levitin
Metal rock
Killing in The
Name
Rage Against
the Machine
Rage Against the
Machine
Produced by Garth Richardson
Contemporary
Diffraction
Yoshihisa Taïra
Quatuor Ixtla
Produced by Amandine Pras
Orchestra
Symphonie #5
Gustav Malher
Wiener Philharmoniker
directed by Pierre Boulez
Deutsche Grammophon
Opera
Lascia ch'io
pianga
George F.
Handel
Not listed
Anechoic recording by Angelo
Farina, downloaded from
www.angelofarina.it
Table 1. Detail of the five musical excerpts used in the study
The pop and opera clips were previously used by
Salimpoor [5], and the contemporary clip was used by
Sutherland [8]. The five CD quality clips were
subjectively matched in loudness by two of the
participants in a preliminary listening test (by applying
attenuation to the louder clips). We encoded the five
excerpts as mp3 with the L.A.M.E. encoder
(lame.sourceforge.net) in order to be able to compare
our findings with those obtained by Salimpoor [5] and
Sutherland [8]. Each excerpt was encoded at five
different bitrates (96, 128, 192, 256 and 320 kb/s),
resulting in six different versions per excerpt. The
experiment consisted of 150 trials corresponding to all
possible pairwise combinations of the six different
versions of the five musical excerpts.
2.3. Procedure
In each trial, participants were asked to listen to both
versions as many times as needed and to choose the
version they preferred in a double blind A/B comparison
task. Each pair was presented twice in counterbalanced
order to nullify order effects. The order of presentation
across trials was randomized. The duration of the
experiment ranged between 60 and 90 minutes per
participant, including a break in the middle of the
experiment. The experiment took place in the Critical
Listening Laboratory of the Centre for Interdisciplinary
Research in Music Media and Technology (CIRMMT,
Montréal, QC, Canada). This ITU standard room
provides high quality controlled listening conditions.
Stimuli were presented through a digital analog
convertor and monitor controller Grace m906 (Grace
Design, Boulder, CO, USA), a Classé CA-5200 stereo
amplifier (Classé Audio, Lachine, QC, Canada) and
B&W 902D loudspeakers (Bowers & Wilkin, Worthing,
West Sussex, England).
2.4. Questionnaire
After the listening test, the participants were asked to
fill out a three-part questionnaire. The first part included
two open (free response) questions asking how difficult
they found the test and how they describe the
differences between two different versions on an
excerpt. We analyzed these open questions using the
constant comparison technique [2]. In the second part,
we investigated which sound criteria the listeners used
to make their decision. For each excerpt, they were
provided with a list of seven sound criteria: High
frequency artifacts, Reverberation artifacts, Dynamic
range, Stereo image, General distortion, Background
noise and Transient artifacts. These criteria were
derived from Sutherland’s [8] collected on expert
listeners who were asked after the listening test to
answer an open question about the sound criteria they
used to discriminate the files in CD quality over mp3.
Participants were also invited to explain, comment or
add other useful sound criteria. Furthermore, we asked
if they were familiar with the musical genre of the clip.
The last part of the questionnaire concerned
demographic information, musical training, studio
experience and listening habits.
Pras et al.
Subjective evaluation of mp3 compression
AES 127th Convention, New York, NY, USA, 2009 October 9–12
Page 4 of 7
Figure 1. Results collapsed over all participants and musical excerpts, grouped by comparison pair: significant
preferences are displayed in grey, non-significant in white.
3. RESULTS
3.1. Overall preference
Over all pairwise comparisons, participants preferred
the higher quality version 68% of times. Figure 1
presents the results as percentage of times participants
preferred the higher quality version, grouped by
comparison pair (results collapsed over all participants
and all excerpts). Using the binomial test, performance
ranging between 46 and 54% is not significant (p>0.05),
meaning that listeners had no significant preference for
one version over the other, which strongly suggests that
they cannot discriminate between the two. Performance
over 54% indicates that listeners could discriminate
between the two versions and that they preferred the
higher quality. Under 46%, the results indicate that
listeners could discriminate between the two versions
and they preferred the lower quality version. Over all
musical excerpts, listeners significantly preferred
(p<0.05) CD quality files to mp3 files for bitrates
ranging from 96 to 192 kbits/s. The results are not
significant between CD quality files and mp3 files for
higher bitrates (256 and 320 kbits/s). Regarding
comparisons amongst mp3 files with different levels of
compression, listeners always significantly preferred the
higher quality version, except for the comparison
between 320 and 256 kbits/s where the results did not
reach statistical significance. These overall results
demonstrate that mp3 compression does introduce
audible artifacts.
3.2. Effect of musical genre and expertise
Our next research question was whether listeners’
sensitivity to these audible artifacts is a function of
musical genre and listeners’ expertise. To test this
hypothesis, we compared the distribution of preferences
using the chi-square statistic on occurrences (counting
the number of times the higher quality version was
preferred, and the number of times the lower quality
version was preferred). The chi-square test revealed a
significant difference between the results’ distribution
amongst the five musical excerpts (χ
2
(4)=22.52,
p<0.001), indicating a highly significant effect of
musical genre on preference. However, there was no
significant difference between the results’ distribution
Pras et al.
Subjective evaluation of mp3 compression
AES 127th Convention, New York, NY, USA, 2009 October 9–12
Page 5 of 7
for pop and metal rock excerpts (χ
2
(1)=0.06, p=0.81
n.s.) so we combined them under Electric clips.
Similarly, there was no significant difference between
the results’ distribution for contemporary, orchestra and
opera excerpts (χ
2
(2)=4.55, p=0.103 n.s.) so we
combined them under Acoustic clips. Finally, a chi-
square test confirmed a significant difference between
the results’ distribution for Electric and Acoustic clips
(χ
2
(1)=17.22, p<0.001), so we present the results for
Electric and Acoustic clips separately.
Figure 2. Results’ distribution into musical genre and
listeners' expertise
A chi-square test revealed a significant difference
between the results’ distribution amongst the four initial
groups of listeners’ expertise: musicians, professional
sound engineers, sound engineer students and research
scientist (χ
2
(3)=53.69, p<0.001), indicating a highly
significant effect of expertise on preference. However,
there was no significant difference between professional
sound engineers and sound engineering students
(χ
2
(1)=1.29, p=0.26 n.s.). There was no significant
difference either between professional sound engineers,
sound engineering students and the research scientist
(χ
2
(2)=2.2, p=0.33 n.s.), while there was a significant
difference between the musicians and the research
scientist (χ
2
(1)=8.8, p<0.01). Therefore, we combined
the results into two expertise levels: one including
professional sound engineers, sound engineering
students and research scientist, referred to as Sound
engineers, and one Musicians group. Here, a significant
difference between the Musicians and Sound engineers
groups was observed (χ
2
(1)=44.27, p<0.001), so we
present the results separately for each group.
Figure 3 presents the percentage of times Musicians
preferred the CD quality files to mp3 files, for every
levels of compression (bitrates). Figure 4 presents the
percentage of times Sound engineers preferred the CD
quality to mp3, for every levels of compression
(bitrates).
Figure 3. Musicians' preference between CD quality and
different levels of compression
Figure 4. Sound engineer's preference between CD
quality and different levels of compression
3.3. Sound criteria
The first part of the questionnaire consisted of two
open-ended questions asking participants to describe the
differences perceived between the different versions,
and the criteria they used to make their judgments. From
the free-format responses, we extracted 34 phrasings
(mean: 2.6 per participant, S.D. 1.8). These phrasings
were assigned to ten non-overlapping categories, seven
of which corresponded to the sound criteria proposed in
the second part of the questionnaire: High frequency
artifacts (6 occurrences), General distortion (5 occ.),
Spatial artifacts - that we named Reverberation - (5
occ.), Clarity of the attacks - that we named Transient
artifacts - (5 occ.), Stereo image (4 occ.), Dynamic
range (2 occ.) and Background noise (1 occ.). Three
additional other categories emerged from the free-
format responses, namely Liveliness (3 occ.),
Articulation of the musical discourse (2 occ.) and Bass
artifacts (1 occ.).
Pras et al.
Subjective evaluation of mp3 compression
AES 127th Convention, New York, NY, USA, 2009 October 9–12
Page 6 of 7
The second part of the questionnaire consisted of close-
ended questions with free-format justification. For each
clip, participants were provided with the list of seven
sound criteria derived from previous research and asked
whether or not they used each criterion for this given
clip, and if so, how and when. Figure 5 shows the
distribution of responses expressed in percentage of the
total number of sound criteria selected by participants.
High frequency artifacts were the most selected
criterion, used for all musical excerpts (86%), while
Background noise was the least selected (28%). It
should be noted that Background noise was not selected
for the metal rock clip at all. This can be explained by
the fact that the original version (CD quality) of this
excerpt is already very noisy (likely because of the
guitar amplifiers and effects). Similarly, Dynamic range
was rarely selected for the pop clip, likely due to the
fact that this musical excerpt sounds extremely
dynamically compressed.
Figure 5. Sound criteria reported by participants to
make their decision during the listening test
These results confirm that trained listeners can indentify
and verbalize the audible artifacts introduced by mp3
compression. However, the distribution of these artifacts
does not vary significantly across musical genres
(χ
2
(6)=22.55, p=0.92 n.s. with Yates’ correction).
3.4. Difficulty of the listening test
An additional open question asked participants how
difficult they found the task. Three reported that was
very difficult, one reported that was not too difficult and
nine reported that was moderately difficult. Over all
musical genres, participants reported that it was hard to
focus (1 occ.) and hard to choose between 2 degraded
files (1 occ.); the differences between two versions
could be very subtle (4 occ.); two versions of the
Acoustic clips (mainly contemporary and opera) were
more difficult to discriminate than two versions of the
Electric clips (6 occ.). We measured familiarity with the
musical genre in the post-questionnaire. Although two
participants reported having difficulty with the Acoustic
clips because they had never listened to these musical
genres, we did not find significant differences between
the overall results and the results for familiar genres.
However, one musician, a professional drummer,
performed significantly better with the pop excerpt
(including clear high-hat beats), as opposed to the four
others excerpts (χ
2
(1)=4.13, p=0.042).
4. DISCUSSION
Together, these results indicate that mp3 compression
introduces audible artifacts, and that listeners’
sensitivity to these artifacts varies as a function of
musical genre and listeners’ expertise. Specifically, we
observed that trained listeners can discriminate and
significantly prefer CD quality over mp3 compressed
files for bitrates ranging from 96 to 192 kbits/s.
Regarding higher bitrates (256 and 320 kbits/s), they
could not discriminate CD quality over mp3 while
expert listeners, with more years of studio experience,
could in the same listening conditions in Sutherland’s
study [8]. Differences between young sound engineers
and experts can be attributed to improved critical
listening skills based on individual listening
experiences. Furthermore, sound engineers and
musicians may not focus on the same sound criteria
when listening to music. While sound engineers are
trained to hear sound artifacts in general situations, the
results from the single case of the professional
drummer, who performed significantly better with the
excerpt including drums than with the four other
excerpts, suggest that musicians are more sensitive to
sound fidelity in specific situations. A future study
testing professional musicians with excerpts including
their personal instrument would allow us to measure
audio formats’ impact on sound fidelity.
An interesting finding is that the artifacts introduced by
mp3 compression were more easily audible on Electric
clips (pop and rock, using amplified instruments) than
on Acoustic clips (using traditional acoustic
instruments). This finding may seem counter-intuitive,
as mp3 compression is used predominantly for popular
music and less frequently classical music. In the sound
engineering community, mp3 format is informally
known to require more headroom (difference between
Pras et al.
Subjective evaluation of mp3 compression
AES 127th Convention, New York, NY, USA, 2009 October 9–12
Page 7 of 7
the peak level of the audio signal and the maximum
possible level to be quantified) than CD format,
although no formal studies on this topic have been
identified. Thus, the different results across musical
genres could be explained by the use of dynamic
compression, often more prevalent in electric music
than in acoustic music.
Although listeners’ sensitivity to mp3 compression is a
function of musical genre, the audible artifacts
introduced by mp3 don’t depend on musical genre. For
all excerpts, High frequency artifacts was reported as
the most easily distinguishable artifact, which concurs
with theoretical errors in high frequency resolution.
Surprisingly, Background noise has been reported as the
least distinguishable artifact, even for acoustic clips that
have a large dynamic range. Although the reduction of
available bits to code the signal intensity is the main
technical consequence of audio compression, the noise
introduced by the quantification approximation is less
perceptible than frequential, temporal and spatial
artifacts. Listeners also reported perceptual changes in
dynamic range, which can be explained by the reduction
of available bits.
Together, these findings suggest that the mastering
process of music productions should be adapted to take
into consideration the common use of mp3 or other
compressed formats. Although the transmission rates of
Internet and the storage devices have greatly evolved
since the introduction of mp3, the use of compressed
formats may continue in early future. This continual
need for compressed formats is enforced due to the
restrictions of space, time and cost to exchange large
amounts of digital information. Furthermore, J. Berger
reported an informal study where young listeners
preferred compressed formats to CD quality [3].
Therefore, there is a need for mastering processes
specifically adapted to lossy compression.
In future studies, we will extend this line of research to
quantify the impact of listening conditions on perceptual
differences between CD quality and mp3 compression.
Furthermore, we will investigate perceptual differences
between CD quality and higher quality formats.
5. ACKNOWLEDGEMENTS
The work reported herein was funded in by an FQRSC
team grant on the Perception of Audio Quality (P.I.: I.
Fujinaga, CG and DJL co-applicants) and a grant from
the Grammy Foundation to DJL. Furthermore, the
authors would like to thank Mark Nelson for his
comments on earlier drafts, as well as Julien Boissinot,
Yves Methot and Harold Kilianski for technical
assistance during the experiments conducted at the
Centre for Interdisciplinary Research in Music Media
and Technology (Montréal, QC, Canada).
6. REFERENCES
[1] Brandenburg, K. (1999). MP3 and AAC explained.
Paper presented at the 17th International
Conference: High-Quality Audio Coding.
[2] Glaser, B. G. (1967). The Discovery of Grounded
Theory: Strategies for Qualitative Research (p.
271). Chicago: Aldine Pub. Co.
[3] Jamieson, A. (2009). iPod generation prefers tinny
music to CD. Retrieved July, 31, 2009, from
http://www.telegraph.co.uk/technology/apple/4941
506/iPod-generation-prefers-tinny-music-to-
CD.html.
[4] Ruzanski, E. P. (2006). Effects of MP3 encoding on
the sounds of music. Institute of Electrical and
Electronics Engineers, 25(2), 43-45.
[5] Salimpoor, V. (2006). Subjective evaluation of
popular audio compression formats. Unpublished
manuscript. McGill University.
[6] Shlien, S. (1994). Guide to MPEG-1 audio
standard. Institute of Electrical and Electronics
Engineers, 40(4), 206-218.
[7] Soulodre, G., Grusec, T., Lavoie, M., & Thibault,
L. (1998). Subjective evaluation of state-of-the-art
two-channel audio codecs. Journal of the Audio
Engineering Society, 46(3), 164-177.
[8] Sutherland, M. E. (2007). The evaluation of
different compression formats: From amateur to
professional listeners. Unpublished manuscript.
McGill University.
Data provided are for informational purposes only. Although carefully collected, accuracy cannot be guaranteed. The impact factor represents a rough estimation of the journal's impact factor and does not reflect the actual current impact factor. Publisher conditions are provided by RoMEO. Differing provisions from the publisher's actual policy or licence agreement may be applicable.
-
- "A single, consistent definition for quality has not yet been offered, however, for certain restricted circumstances, 'quality' has an understood meaning when applied to audio. Measurement techniques exist for the assessment of audio quality, such as [1] and [2], however such standards typically apply to the measurement of quality with reference to a golden sample; what is in fact being ascertained is the reduction in perceived quality due to destructive processes, such as the effects of compression codecs, in which the audio being evaluated is a compressed version of the reference and the deterioration in quality is measured [3]. Such descriptions would not strictly apply to the evaluation of quality in musical recordings. "
Conference Paper: Perception & Evaluation of Audio Quality in Music Production
[Show abstract] [Hide abstract]
ABSTRACT: A dataset of audio clips was prepared and audio quality assessed by subjective testing. Encoded as digital signals, a large amount of feature-extraction was possible. A new objective metric is proposed , describing the Gaussian nature of a signal's amplitude distribution. Correlations between objective measurements of the music signals and the subjective perception of their quality were found. Existing metrics were adjusted to match quality perception. A number of timbral, spatial, rhythmic and amplitude measures, in addition to predictions of emotional response, were found to be related to the perception of quality. The emotional features were found to have most importance, indicating a connection between quality and a unified set of subjective and objective parameters. -
- "It is unlikely that this significantly affected our data. Across a large number of unrelated, independent experiments, involving a range of listeners and material, we have found no difference in listening judgments obtained from using headphones versus loudspeakers, provided that both are of sufficiently high quality, such as those used in the present experiments (Pras, Zimmerman, Levitin, & Guastavino, 2009;Salimpoor, Guastavino, & Levitin, 2007). An interesting difference between the Varied Timing and Varied Amplitude conditions is that, in the Varied Amplitude condition, random-A was rated as more emotional than mechanical and no different from expressive-A. "
[Show abstract] [Hide abstract]
ABSTRACT: Expression in musical performance is largely communicated by the manner in which a piece is played; interpretive aspects that supplement the written score. In piano performance, timing and amplitude are the principal parameters the performer can vary. We examined the way in which such variation serves to communicate emotion by manipulating timing and amplitude in performances of classical piano pieces. Over three experiments, listeners rated the emotional expressivity of performances and their manipulated versions. In Experiments 1 and 2, timing and amplitude information were covaried; judgments were monotonically decreasing with performance variability, demonstrating that the rank ordering of acoustical manipulations was captured by participants' responses. Further, participants' judgments formed an S-shaped (sigmoidal) function in which greater sensitivity was seen for musical manipulations in the middle of the range than at the extremes. In Experiment 3, timing and amplitude were manipulated independently; timing variation was found to provide more expressive information than did amplitude. Across all three experiments, listeners demonstrated sensitivity to the expressive cues we manipulated, with sensitivity increasing as a function of musical experience. -
- "Previous audio quality research in the context of digitization has explored listeners' ability to discriminate between audio files presented in CD audio and compressed MP3 formats [1]. Pras and Guastavino [2] have also compared the differentiability of sample rates. "
[Show abstract] [Hide abstract]
ABSTRACT: Digitization of phonograph records is an important step towards the preservation of our cultural history and heritage. The phonograph playback systems (PPS) required for this digitization process are comprised of several components in a variety of price ranges. We report on the results of two listening tests intended to ascertain the extent to which expert listeners can discriminate between PPS of different price ranges. These results are intended to determine the extent to which component selection affects the discrimination between PPS and to provide a set of guidelines for the purchase of PPS components for the digitization of phonograph record collections.
