Investigation of various algorithms on multichannel audio compression

Full text

Proc. of the 4th IEEE International Conference on Smart Instrumentation, Measurement and Applications (ICSIMA)
28-30 November 2017, Putrajaya, Malaysia
Investigation of Various Algorithms on
Multichannel Audio Compression
Teddy Surya Gunawan#, Siti Aisyah Abdul Rashid#, Mira Kartiwi*
#Electrical and Computer Engineering Department, International Islamic University Malaysia
*Information Systems Department, International Islamic University Malaysia
Corresponding email: tsgunawan@iium.edu.my
second.com
Abstract — Multichannel audio or surround sound
compression is rather more challenging to compress compare to
mono and stereo audio. Nowadays, many methods and
algorithms have been proposed to improve the compression
performance on multichannel audio. This book focuses on
performance evaluation of various algorithms on multichannel
audio compression. First, we identified and investigated current
state-of-the-art audio compression algorithms, both lossless and
lossy compression, which can handle mono, stereo, 5.1, and 7.1
multichannel audio. Out of various algorithms available, AC3,
AAC, and Ogg have been selected as lossy compression
algorithms, while FLAC and MPEG-4 ALS have been chosen as
lossless compression algorithms. Two performance measure were
used in the experiments, i.e. compression ratio and encoding
time. The results showed that among three lossy audio
compression algorithms, AC3 has the fastest encoding time while
Ogg Vorbis has the highest compression ratio. Furthermore,
between FLAC and MPEG-4 ALS, FLAC has faster encoding
time and MPEG-4 ALS has higher compression ratio. Overall, in
terms of encoding time and compression ratio, it has been found
that FLAC is the fastest coder while Ogg Vorbis has the highest
compression ratio among five encoders evaluated.
Keywords—multichannel audio; lossless compression; lossy
compression; encoding time; compression ratio.
I. INTRODUCTION
Multichannel audio systems are widely used in modern
sound devices. Usually, two digits separated by a decimal
point, e.g. 2.1, 4.1, 5.1, 6.1, 7.1, are used to classify the various
kinds of speaker set-up [1, 2]. This number represents the
number of audio tracks used. Some audio systems only have a
single channel or two channels (stereophonic sound or 2.0
channel sound). The first digit shows the number of primary
channels, i.e. satellite units, each of which are reproduced on a
single speaker which has the capability to handle range of
frequency between 100Hz to 22 kHz. On the other hand, the
second digit (decimal digit) represents the presence of LFE
(Low Frequency Effect) that is reproduced on a subwoofer.
Moreover, surround system describes a type of audio output in
which the sound appears to surround the listener by 360
degrees, in which it gives impression that sound are coming
from all possible directions. It has been used to provide a more
realistic and engaging experience [3].
There are two kinds of audio compression algorithm those
are lossy and lossless. Lossy audio compression is known by
their well-designed system to shrinks file sizes. Advanced
Audio Coding (AAC), MPEG-1 Layer III (MP3), Dolby AC-3,
Opus, OGG Vorbis [4] and Windows Media Audio Lossy
(WMA lossy) are the examples of prevalent foremost lossy
audio coding system [5]. AAC can be considered as the most
influential multichannel audio coding algorithm [6]. This is due
to its ability to support audio channels up to 48 channels and
contribute lossless audio for 5.1 channels at sampling rates 320
kbits/s. Meanwhile, AC3 provides high audio quality at
384kbit/s [7].
Meanwhile, the most well-known codec in lossless
algorithm are Free Lossless Audio Codec (FLAC), Apple
Lossless Audio Codec (ALAC), Waveform Audio File (WAV),
MPEG-4 Audio lossless [8], True Audio (TTA) [9]. Each of
the codec, have the own domain and advantage to encode and
decode the audio. Lossless methods do not have any loss
information and provide an exact replica of the original signal.
Although many research has been conducted on lossless
and lossy audio compression, but not many researches have
been focused on the multichannel audio coding. Therefore, the
objective of this paper is to investigate the performance of
various audio compression algorithms to encode multichannel
audio in terms of encoding time and compression ratio.
II. MULTICHANNEL AUDIO
A. Monoaural and Stereophonic Audio
Fig. 1. Stereo Speakers Setup [10]

From analog audio, sampling and quantization are
conducted to represent the sound wave into digital
representation. A stereo signal can be considered as two
independent channels of audio information, i.e. left and right
channels. Stereophonic audio provides the impression of sound
localization [10]. Fig. 1 illustrates the stereo setup in a typical
living room.
B. 5.1 Multichannel Audio
Unlike mono and stereo audio, multi-channel audio format
designates in more than two channels. This type of audio
format aims to advance the ability of sound localization. As an
example, a 5.1 multichannel loudspeakers arrangement has
been illustrated in Fig. 2. The left and right channels placed at
±30˚ like in stereo audio. Meanwhile, the rear right and left
channel located at ±110˚. Usually, they are used for extended
sound source localizations interpretation. For center channel, 0˚
commonly for playing again voice contents in moving audio.
The decimal digit (.1) channel refer to subwoofer channel
which also recognize as LFE channel. This channel is for
playing back the low frequency contents. By adding more
surround loudspeaker to the two standard channels LS and RS,
it will create larger listening zone. This setup had been widely
used in cinema [10].
Fig. 2. 5.1 Multichannel Speakers Setup [10]
C. 7.1 Multichannel Audio and Beyond
Multichannel audio 7.1 is a further enhancement to 5.1
audio channels. There are other two side-surround speaker in
the speaker configuration. Many of application used 7.1 audio
in order to greater impact of surround sound. The loudspeakers
arrangement is almost similar to 5.1 multichannel audio.
However, there are another two speaker left and right rear
which about ±135˚to surround sound. Fig. 3 shows the setup
configuration of multichannel 7.1 audio.
Beyond 7.1 multichannel audio, 10.2 channel surround
sound has been developed. It is the advanced version of 5.1
technology, but 10.2 could produce twice as good as 5.1. In this
channel configuration. 14 channels are used to including five
front speakers, five surround channels, two LFE and two
heights, plus the addition of a second sub-woofer [10].
Fig. 3. 7.1 Multichannel Speakers Setup [10]
III. AUDIO COMPRESSION ALGORITHMS
Of the various lossless and lossy compression algorithms
available, Dolby AC3, AAC and Ogg Vorbis have been
selected as lossy compression algorithms, while FLAC and
MPEG-4 ALS has been chosen as lossless compression
algorithms. It is selected due to its capability to encode
multichannel audio and its popularity.
A. Dolby AC3 Encoder
AC3 is Dolby Audio Codec 3/Advanced Codec 3/Acoustic
Coder 3 which refers to multichannel compression technology
that has been developed by Dolby Laboratories. The objective
of this codec is to compress audio as similar as possible to the
original signal while using a minimum bit rate. AC3 had been
practice at cinema due to its outstanding sound system. Fig. 4
shows Dolby AC3 encoder mechanisms.
Fig. 4. Dolby AC3 Encoder
At the AC3 encoder process, the algorithm will use MDCT
in audio transformation from time to frequency domain. Then,
transform coefficients will be grouped into non uniform
subbands. The subbands are approximately the critical bands of
human auditory system. From that, transform coefficients
within one subband are converted to a floating-point

representation, with one or more mantissas per exponent. The
exponents are encoded by a suitable strategy according to time
and frequency resolution and then go into the psychoacoustic
model. In psychoacoustic model, the perceptual resolution is
calculates according to the encoded exponents and the proper
perceptual parameters.
B. Advanced Audio Coding (AAC)
AAC leads MP3 as there is a new non-backward
compatible audio coder introduced in [1, 6]. It becomes popular
due to application in Apple iTunes. Fig. 5 illustrates an AAC
encoder. AAC operates MDCT transform only in its main
coding loop and transient detection function to detect a long
window of 2048 points or a serial set of eight 256 point
windows is ready for the MDCT transform. Thus, this give
high frequency resolution of 23Hz and 2.7ms for a signal
sampled at 48 kHz. A gain control procedure is incorporated in
the SSR profile of AAC. A Pseudo Quadrature Mirror Filter
(PQMF) filter bank is used to split the signal into four
subbands with same bandwidth. The original signal sampling
rates reduced to quarters by discarding one or more subbands.
AAC utilizes the temporal-noise-sharping technique to expel
the pre-echo effect caused by transients. Based on subjective
evaluations, AAC provides great audio for 5 channel
bandwidth at bit rate of 320kbps.
Fig. 5. AAC encoder
C. Ogg Vorbis
Ogg Vorbis is a full open, non-proprietary, patent and
royalty free compression audio format. Fig. 6 illustrates the
possible implementation of Ogg Vorbis encoder. It is based on
vector quantization and transformation with overlapping
windows, i.e. modified discrete cosine transform (MDCT).
Each windows can have 2048 or 512 samples. The shorter one
is used only to encode a transient signals. After transformation
to frequency domain, the signal is analyzed by psychoacoustic
model and inaudible part of the spectrum is removed. Then the
floor vector is generated for each of the channels.
Fig. 6. Ogg Vorbis encoder
D. Free Lossless Audio Coder (FLAC)
Free Lossless audio coding (FLAC) is quite famous among
lossless codec due its fastest decoding audio. FLAC uses a
linear prediction mathematical (LPC) operation where future
values of the digital signal are estimated as a linear function of
previous samples. The FLAC encoder first divide the input
audio signal into frames. Then, it will conduct an interchannel
decorrelation. The predictor is then attempts to find an
optimum coefficients to predict the signal. Lastly, the predictor
coefficients and its residue were passed to entropy coding.
E. MPEG-4 Audio Lossless Coding (ALS)
MPEG-4 Audio Lossless Coding (ALS) standard is derived
from MPEG-4 audio coding standard. This codecs feature is to
preserve every single bit of the original audio data. ALS
provides method for lossless coding of audio signals with
arbitrary sampling rates, resolutions of up to 32-bit and up to
216 channels, also including 32-bit floating-point signals.
Thus, virtually all known input formats from CD quality 44.1
kHz, 16-bit to high-end audio multichannel can be supported.
Fig. 6 illustrates the ALS encoder.
Fig. 6. MPEG-4 ALS encoder
IV. RESULTS AN DISCUSSION
In this section, the audio encoders implementation, audio
database, as well as performance evaluation in terms of
encoding time and compression ratio will be conducted. To
simplify the experiments, this section will focus only on 5.1
and 7.1 multichannel audio.
A. Implementation
Table I shows the audio encoder along with its software
implementation. NeroAACEnc version 1.5.4.0 release in
February 2010 was used for AAC encoding. Oggenc2 v2.88
(libvorbis 1.3.5) was used for Ogg encoding. Finally, FFmpeg
was used for AC3, FLAC, and MPEG-4 ALS encoding.
TABLE I. AUDIO ENCODER AND ITS IMPLEMENTATION
Algorithm Audio Format Extensions Software
Lossy
AAC .aac NeroAACEnc
AC3 .ac3 FFmpeg
OGG Vorbis .ogg oggenc2
Lossless FLAC .flac FFmpeg
MPEG-4 ALS .m4a FFmpeg
Fig. 7 shows the flowchart of the overall implementation
using Matlab. The program will loop through five encoders, i.e.
AAC, Ogg, AC3, FLAC, and MPEG-4 ALS, and five files
each for mono, stereo, 5.1, and 7.1 multichannel audio signals.

For each file and each encoder, the compression ratio is
recorded. System call, i.e. dos() function in Matlab, is
employed to access the encoder executable file. For accuracy,
the program will loop 10 times for each coder, in which the
average value of encoding time will be recorded.
Fig. 7. Overall Implementation using Matlab
B. Audio Database
Various audio signal in the original WAV format are
collected from internet. Table II shows the audio database used
for experimentation on 5.1 and 7.1 multichannel audio.
TABLE II. AUDIO DATABASE FOR MULTICHANNEL AUDIO
Channel File Name Details
5.1
Five1.wav 47 seconds, 44100 Hz
Five2.wav 9 seconds, 48000 Hz
Five3.wav 131 seconds, 44100 Hz
Five4.wav 300 seconds, 44100 Hz
Five5.wav 125 seconds, 44100 hz
7.1
Seven1.wav 32 seconds, 48000 Hz
Seven2.wav 14 seconds, 48000 Hz
Seven3.wav 95 seconds, 48000 Hz
Seven4.wav 19 seconds, 48000 Hz
Seven5.wav 4 seconds, 48000 Hz
C. Experiments on Lossy Compression
The AAC, AC3 and OGG are encoder that will compress
the original audio WAV format which is lossless to lossy
format. As the encoding is conducted in Matlab, the output
from each type of audio is shown in Table III and IV. The
columns ‘T_AAC’, ‘T_OGG’ and‘T_AC3’ indicate the
encoding time of audio files in AAC, OGG and AC3.
Meanwhile, the columns ‘C_AAC’, ‘C_OGG’ and ‘C_AC3’
specify the compression ratio of audio files in AAC, OGG and
AC3. In order to compare the best time processing and
compression ratio, yellow and blue highlight is used to identify
them. The blue color is for the best compression ratio in each
audio file. On the other hand, the least processing time is
shown by yellow color.
TABLE III. LOSSY COMPRESSION FOR 5.1 MULTICHANNEL AUDIO
5.1 AAC AC3 OGG
Audio T_AAC C_AAC T_AC3 C_AC3 T_OGG C_OGG
Five1 5.488 32.62 0.598 9.454 3.077 21.293
Five2 1.118 43.293 0.203 10.274 0.651 24.731
Five3 8.114 10.243 1.551 9.45 8.458 17.162
Five4 20.056 10.333 3.511 9.45 18.712 18.258
Five5 7.788 10.595 1.45 9.449 8.011 18.467
TABLE IV. LOSSY COMPRESSION FOR 7.1 MULTICHANNEL AUDIO
7.1
Audio
AAC AC3 OGG
T_AAC C_AAC T_AC3 C_A C3 T_OGG C_OGG
Seven1 4.393 20.559 0.591 13.719 3.211 11.502
Seven2 9.336 117.141 0.244 20.546 0.761 229.361
Seven3 20.404 38.271 1.442 20.569 7.132 47.283
Seven4 13.572 55.577 0.315 13.694 1.218 97.631
Seven5 1.78 38.867 0.189 12.558 0.353 91.7
D. Experiments on Lossless Compression
There are two encoders for lossless to lossless compression.
Both MPEG 4 ALS and FLAC had been implementing in
Matlab to synthesis the time processing and compression ratio
between each file. The columns ‘T_FLAC’ and ‘T_M4A’
indicates the encoding time which the file was encoded by
FLAC and MPEG 4ALS. Compression ratio is showed by
‘C_FLAC’ and ‘C_M4A’ column. The blue highlight the
highest compression ratio in each audio file. Meanwhile,
yellow highlight the smallest time processing of audio while
encoding is done by FLAC and MPEG 4 ALS.

TABLE V. LOSSLESS COMPRESSION OF 5.1 MULTICHANNEL AUDIO
5.1 Audio
FLAC MPEG4 ALS
T_FLAC C_FLAC T_M4A C_M4A
Five1 0.468 7.035 5.723 12.337
Five2 0.168 9.715 1.784 13.008
Five3 1.447 3.003 13.141 12.344
Five4 2.855 3.488 31.625 12.328
Five5 1.348 3.366 12.94 12.337
TABLE VI. LOSSLESS COMPRESSION OF 7.1 MULTICHANNEL AUDIO
7.1 Audio
FLAC MPEG4 ALS
T_FLAC C_FLAC T_M4A C_M4A
Seven1 0.474 4.002 7.45 12.454
Seven2 0.266 27.338 1.34 93.798
Seven3 1.764 4.418 21.677 23.488
Seven4 0.292 30.142 3.301 39.292
Seven5 0.142 8.437 0.598 21.469
E. Discussion
All the lossy and lossless encoders had been evaluated in
Matlab to differentiate which one is the best coder to encode
audio in terms of encoding speed (time processing) and
compression ratio. There are 100 files are being encoded in
AAC, AC3, OGG (lossy compression), and FLAC, MPEG-
ALS (lossless compression).
Table III to IV shows the results in lossy compression. The
evaluation is conducted by signify which encoder of each audio
has the possibility to have smallest result in encoding time
represent in yellow and biggest compression ratio represent in
blue. As we compare them in each file, the average best results
seem goes to AC3 encoder for encoding time. Around 100% of
audio file have AC3 for the faster processing. These amounts
show AC3 can encode audio file in the fastest way than AAC
and AC3. For compression ratio, OGG encoder compressed the
audio better among other codec. Around 80% audio files have
largest compression ratio on OGG, followed by 20% audio
files on AAC.
Table V and VI shows the results in lossless compression.
The proportion of encoding time is best at FLAC encoder. This
is due to absolute 100% of audio file in m 5.1 and 7.1 audio
have minimum time compared to MPEG4 ALS encoder. In
contrast, the best compression ratio for lossless algorithm is
MPEG 4 ALS encoder as all audio file has largest ratio in the
encoder.
From the observations of lossy encoder comparison, we can
examine that the best encoder for encoding time is AC3.This
encoder give good result in encoding time at 5.1 and 7.1
multichannel audio. AC3 is matured in term of encoding time
compared to other codec. In lossless audio, when the encoders
encode all the audio in 5.1 and 7.1 multichannel audio, the
finding for the best encoding time and compression ratio give a
consistent result. The entire 10 audio file shows similar pattern.
FLAC is the best to encode audio at smallest speed compare to
MPEG-4 ALS. However, in term of compression ratio, MPEG-
4 ALS performs better than FLAC.
V. CONCLUSIONS AND FUTURE WORKS
This paper has presented the performance evaluation of
three lossy and two lossless audio compression evaluated on
5.1 and 7.1 multichannel audio signals. It has been found that
among three lossy audio compression algorithms, AC3 has the
fastest encoding time while Ogg Vorbis has the highest
compression ratio. Furthermore, between FLAC and MPEG-4
ALS, FLAC has faster encoding time and MPEG-4 ALS has
higher compression ratio. Overall, in terms of encoding time
and compression ratio, it has been found that FLAC is the
fastest coder while Ogg Vorbis has the highest compression
ratio among five encoders evaluated. Future works could
include the optimize parameters of each audio compression
algorithm to better evaluate its performance.
ACKNOWLEDGMENT
The authors would like to express their gratitude to the
Malaysian Ministry of Higher Education (MOHE), which has
provided funding for the research through the Fundamental
Research Grant Scheme, FRGS15-194-0435.
REFERENCES
[1] M. Bosi and R. E. Goldberg, Introduction to digital audio coding and
standards, vol. 721, Springer Science & Business Media, 2012.
[2] F. Rumsey, Spatial audio, CRC Press, 2012.
[3] F. Schuh, S. Dick, R. Füg, C. R. Helmrich, N. Rettelbach, and T.
Schwegler, "Efficient multichannel audio transform coding with low
delay and complexity," in Audio Engineering Society Convention 141,
pp., 2016.
[4] J. Moffitt, "Ogg Vorbis—open, free audio—set your media free," Linux
journal, vol. 2001, pp. 9, 2001.
[5] W. Jackson, "Audio Concepts, Terminology, and Codecs," in Android
Apps for Absolute Beginners: Springer, 2014, pp. 651-663.
[6] M. Bosi, K. Brandenburg, S. Quackenbush, L. Fielder, K. Akagiri, H.
Fuchs, and M. Dietz, "ISO/IEC MPEG-2 advanced audio coding," Journal
of the Audio engineering society, vol. 45, pp. 789-814, 1997.
[7] R. Hennequin, J. Royo-Letelier, and M. Moussallam, "Codec independent
lossy audio compression detection," in Acoustics, Speech and Signal
Processing (ICASSP), 2017 IEEE International Conference on, pp. 726-
730, 2017.
[8] T. Liebchen, T. Moriya, N. Harada, Y. Kamamoto, and Y. Reznik, "The
MPEG-4 Audio Lossless Coding (ALS) standard-technology and
applications," in Proc. 119th AES Conv, pp., 2005.
[9] A. Djuric, "TTA Lossless audio codec-True audio compressor
algorithms," 2010.
[10] T. Holman, Surround sound: up and running, CRC Press, 2014.