Citation
Analyzing the long term average sorted spectrum of audio uploaded to YouTube

Material Information

Title:
Analyzing the long term average sorted spectrum of audio uploaded to YouTube
Creator:
Rabbio, Andew Ryan
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Master's ( Master of science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Music and Entertainment Industry Studies, CU Denver
Degree Disciplines:
Recording arts
Committee Chair:
Grigoras, Catalin
Committee Members:
Smith, Jeff M.
Whitecotton, Cole

Notes

Abstract:
This thesis introduces and documents the collection of a general-purpose dataset of audio recordings from several recorders at all available settings. As a pilot study, this dataset was used in a study of YouTube effects on audio recompression. This is accomplished by analyzing the Long Term Average Sorted Spectrum (LTASS) of audio files before and after being uploaded to YouTube. Over 350 recordings are included in the dataset from a variety of recording devices and manufacturers to ensure a diverse dataset. All recordings were made under controlled conditions to ensure the results are easily comparable and reproducible. By analyzing the LTASS of audio uploaded to YouTube this paper will provide greater clarity to the effects YouTube has on audio compression across a variety of device settings.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
Copyright Andrew Ryan Rabbio. Permission granted to University of Colorado Denver to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.

Downloads

This item has the following downloads:


Full Text
ANALYZING THE LONG TERM AVERAGE SORTED SPECTRUM OF AUDIO UPLOADED TO
YOUTUBE
by
ANDREW RYAN RABBIO B.S., University of Colorado Denver, 2016
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Recording Arts Program
2019


©2019
ANDREW RYAN RABBIO ALL RIGHTS RESERVED
11


This thesis for the Master of Science degree by Andrew Ryan Rabbio has been approved for the Recording Arts Program by
Catalin Grigoras, Chair Jeff M. Smith Cole Whitecotton
Date: May 18, 2019


Rabbio, Andrew Ryan (M.S., Recording Arts Program)
Analyzing the Long Term Average Sorted Spectrum of Audio Uploaded to YouTube Thesis directed by Associate Professor Catalin Grigoras
ABSTRACT
This thesis introduces and documents the collection of a general-purpose dataset of audio recordings from several recorders at all available settings. As a pilot study, this dataset was used in a study of YouTube effects on audio recompression. This is accomplished by analyzing the Long Term Average Sorted Spectrum (LTASS) of audio files before and after being uploaded to YouTube. Over 350 recordings are included in the dataset from a variety of recording devices and manufacturers to ensure a diverse dataset. All recordings were made under controlled conditions to ensure the results are easily comparable and reproducible. By analyzing the LTASS of audio uploaded to YouTube this paper will provide greater clarity to the effects YouTube has on audio compression across a variety of device settings.
The form and content of this abstract are approved. I recommend its publication.
Approved: Catalin Grigoras
IV


TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION...........................................1
Audio Compression................................2
FFmpeg...........................................3
YouTube-dl.......................................4
Previous Research................................4
II. YOUTUBE FRAMEWORK......................................6
Available Formats................................6
III. TEST RECORDING FRAMEWORK...............................8
Uploading and Downloading From YouTube..........10
MATLAB Analysis.................................11
IV. RESULTS...............................................12
V. CONCLUSION............................................27
REFERENCES.....................................................29
v


LIST OF TABLES
TABLE
1. Device settings recorded for each recorder...............................9
2. Outline of every file in the dataset along with the CC, MQD, sample rate before/after
download, and the bitrate before/after download
16


LIST OF FIGURES
FIGURE
1. Example of digital handheld recorder .................................2
2. FFmpeg command to combine exemplar recording with test video .........3
3. Example of Youtube-dl query for available formats of a 192kbps mono Mp3 file ..7
4. LTASS Results for "01ympus_WS853_8kbps_MonoMP3_LCoff"..............13
5. LTASS Results for "01ympus_WS853_64kbps_MonoMP3_LCoff..............13
6. LTASS Results for "01ympus_WS853_8kbps_MonoMP3_VSYNC_5sec_03"......14
7. LTASS Results for "Oiympus_WS853_64kbps_MonoMP3_VSYNC_5sec"........14
8. LTASS Results for "01ympus_WS100_LP_MICSENS_high"..................15
9. LTASS Results for "Sony ICD PX333 48kbps Mono MP3 LCon MICSENS med"...15
10. LTASS Results For "Tascam DR07 16bit Stereo Wav 44.1kHz".............16
11. Plot of Bit Rate vs. Correlation Coefficient For All Recordings...23
12. Plot of Bit Rate vs. Correlation Coefficient For Olympus WS100...23
13. Plot of Bit Rate vs. Correlation Coefficient For Olympus WS853...24
14. Plot of Bit Rate vs. Correlation Coefficient For Sony ICD PX333...24
15. Plot of Bit Rate vs. Correlation Coefficient For Tascam DR-07...25
16. Plot of 8kbps 44.1KHz Recordings And Their Corresponding CC Values...25
17. Plot of 128kbps 44.1KHz Recordings And Their Corresponding CC Values.26


LIST OF ABBREVIATIONS
LTASS - Long Term Average Sorted Spectrum
SWGDE - Scientific Working Group on Digital Evidence
MP3 - MPEG-1 Audio Layer-3
WMA - Windows Media Audio
WAV - Waveform Audio File Format
ABR - Adaptive BitRate
DASH - Dynamic Adaptive Streaming over HTTP
HTTP - HyperText Transfer Protocol
MQD - Mean Quadratic Difference
CC - Correlation Coefficient
PCM- Pulse Code Modulation
LP - Long-Term Recording mode on Olympus WS-100
HQ - High Quality recording mode on Olympus WS-100
SP - Standard recording mode on Olympus WS-100
LCon - Low Cut Filter on
LCoff - Low Cut Filter off
MICSENS - Mic Sensitivity setting
VOR - Voice Operated Recording Setting on Sony PX-333
VSYNC - Voice Sync Setting on Olympus WS-853
VCVA - Variable Control Voice Actuator setting on Olympus WS-10


IX


CHAPTER I
INTRODUCTION
YouTube has over 1.9 billion unique users that visit the site each month, watching over 1 billion hours of video content every day making it the most common video streaming service on the Internet [11]. With its widespread use, it is inevitable for videos depicting criminal activity, and confessions of crimes in some cases, to be present. There have been many instances where these recordings have been entered into courtrooms as evidence [12]. With this in mind, it is important to know what is happening to these videos when they are uploaded to YouTube and then downloaded using services such as Youtube-dl or other third party programs.
There have been previous publications analyzing the effects of YouTube on video compression but research is lacking for the analysis of the audio from these videos, which is where this paper looks to expand. With the growing popularity of handheld recorders such as small Olympus and Tascam devices being used by many governmental and private investigators as well as public consumers, these devices were used to make the exemplar recordings that comprise the dataset (Figure 1). Most of these devices do not record in a proprietary format, and instead record in standard formats such as MP3, WAV, and WMA, which poses problems for forensic investigators. These standard file formats are much easier to manipulate and re-encode into different formats making it harder to know if a recording is original.
This paper will provide information on the Long Term Average Sorted Spectrum of exemplar recordings taken from 4 different devices from popular manufacturers and compare those results to the LTASS of the recordings after they have been uploaded and
1


downloaded from YouTube. By analyzing the differences between the original recordings and the uploaded/downloaded from YouTube versions, a clearer concept of how YouTube compresses audio from a variety of different file formats and device settings will be gained
Figure 1 Example of digital handheld recorder (Olympus.co.uk)
Audio Compression
While many in the public may not realize, audio compression is in use all around us every day from the music we listen to, what we hear and see on TV, to the sounds that are emitted from a child’s favorite toy. The advent of audio compression began in the 1980’s with the need to reduce the bit-rate requirement for Compact Disk (CD) without sacrificing noticeable decreases in audio quality. Since then a variety of audio compression algorithms, commonly known as codecs, have been introduced that exploit the way we hear to reduce the size of files. The main way that these codecs do this is called perceptual coding [7]. During this process the compression algorithm looks to exploit imperfections of
2


human hearing with the goal of making the file size smaller while maintaining sound quality. Perceptual coding is a lossy compression technique that utilizes temporal and simultaneous masking to eliminate information that our ears are unable to hear. By eliminating this information and represent the digital audio signal with the least about of bits while maintaining transparent signal production.
FFmpeg
FFMPEG is a powerful command line tool that allows for the conversion, creation, and streaming of digital video and audio files. It is offered for free from the developers website and allows the user to utilize its built in libraries. The FFmpeg framework is utilized in a variety of application for forensic and commercial purposes.
Audio by itself cannot be uploaded to YouTube and must be accompanied by some video component which is why for the experiments detailed in this paper FFmpeg was used to mux the exemplar recordings with a static video. Figure 2 below gives an example of the command used to combine the exemplar recordings with the test video, which is named "black.mp4". This command stores the original audio as an uncompressed 16bit pcm audio stream in the output
Andrews-MacBook-Pro-2:research andrewrabbio$ ffmpeg -i black.mp4 -i Olympus_WS10 0_HQ_MICSENS_high_2.WMA -c:v copy -c:a pcm_sl6le -shortest Orig-Olympus_WS100_HQ _MICSENS_high_2_Upload.avi|
Figure 2 FFmpeg command to combine exemplar recording with test video
YouTube-dl
Youtube-dl is a free cross-platform command line tool that allows the user to download video and audio from YouTube. The use of this tool is considered best practice compared
3


to other third party programs that purport to download videos from YouTube as it allows for the download of all available file types and resolutions from YouTube’s servers compared to other tools, which offer a limited download options. For the tests described in this paper this tool was used to download the audio from videos that were uploaded to YouTube. For the purposes of this test being more reproducible the account that was used to upload the videos was not used when the videos were downloaded using Yoube-dl.
Previous Research
Currently because of the nature of YouTube being a video streaming service not dedicated to audio the bulk of research relating to this topic has emphasized the authentication of videos that have been re-encoded by YouTube, and source identification of high definition videos. In the Paper "YouTube Re-Compression Effects" Witecotton performs a variety of tests to figure out what information can be gleaned from files that are uploaded to YouTube using third-party options to analyze the files. In this paper he discusses the encoding algorithm that YouTube uses to re-encode videos. He arrives at the conclusion that regardless of the original container type the video stream data remains the same after download from YouTube’s server using a variety of different third party applications such as youtube-dl, i.e. the algorithm is consistent across multiple containers [3],
In the paper "Source Identification of High Deffinition Videos: A Forensic Analysis of Downloaders and YouTube Video Compression Using a Group of Action Cameras" Giammarrusco gives a thorough explanation of YouTube’s framework and how it re encodes videos which was integral to the research done in this paper. He looks at a series of third party downloader tools, which at the time of publication were most popular when
4


it came to downloading videos from YouTube, analyzing the resulting structure and metadata after each tool was used to download a video from YouTube. This analysis lead to the understanding that these tools and the re encoding that YouTube automatically does manipulated much of the metadata information in the header of the file, which has strong implications for forensic analysis and authentication purposes [5].
5


CHAPTER II
YOUTUBE FRAMEWORK
A firm understanding of YouTube’s framework is essential to understand what is happening when videos are uploaded and downloaded, either from the site or using third party applications. When a video is uploaded to YouTube, regardless of the original format, size, and dimensions, it is automatically re encoded which may have certain ramifications for authentication and analysis purposes. With this is mind it is important to note that the file that is uploaded will be used to create a variety of different video streams at different resolutions. As Gimmarrusco puts it, the file you upload is like a master file that is used to create different video streams, "Simply stated, the better the quality of file that is uploaded, to YouTube, the better quality that will be received upon download" [5]
These different video streams are created using adaptive dynamic streaming over HTTP (DASH) during playback, which is an Adaptive BitRate (ABR) video streaming technology. ABR allows YouTube to switch the video and audio quality based on the users available connection. Anorga et al. explains that "The main outcome of this feature is that if on YouTube's player quality parameter is set on ‘auto’, YouTube can adapt the bitrate of the video based on the client’s available download bandwidth, so the video streaming is adapted to a dynamic environment." [9].
Available Formats
In order to see the different audio qualities that YouTube offers during playback based on the users available bandwidth youtube-dl can be used to query for all available audio and video formats. Figure 3 below shows an example of this query of available formats of a 192kbps mono Mp3 file.
6


[info] Available formats for gQQ6pqX6zXc:
format code extension resolution note
139 m4a audio only DASH audio 49k , m4a_dash container, mp4a.40.5@ 48k (22050Hz)
140 m4a audio only DASH audio 130k , m4a_dash container, mp4a.40.2@128k (44100Hz)
134 mp4 640x360 DASH video 10k , mp4_dash container, avcl.4d4016, 6fps, video only
136 mp4 1280x720 DASH video 19k , mp4_dash container, avcl.4d401f, 6fps, video only
278 webm 256x144 DASH video 95k , webm_dash container. , vp9, 6fps, video only
160 mp4 256x144 DASH video 108k , mp4_dash container, avcl.4d400b, 6fps, video only
242 webm 426x240 DASH video 220k , webmjash container. , vp9, 6fps, video only
133 mp4 426x240 DASH video 242k , mp4_dash container, avcl.4d400c, 6fps, video only
243 webm 640x360 DASH video 405k , webm_dash container. , vp9, 6fps, video only
244 webm 854x480 DASH video 752k , webm_dash container, , vp9, 6fps, video only
135 mp4 854x480 DASH video 1155k , mp4_dash container, avcl.4d4014, 6fps, video only
247 webm 1280x720 DASH video 1505k , webm_dash container. , vp9, 6fps, video only
43 webm 640x360 medium , vp8.0, vorbis@128k, 284.40KiB
18 mp4 640x360 medium 107k , avcl.42001E, mp4a.40.2@ 96k (44100Hz), 348.43KiB
22 mp4 1280x720 hd720 148k , avcl.64001F, mp4a.40.2 Figure 3 Example ofYoutube-dl query for available formats of a 192kbps mono Mp3 file.
7


CHAPTER III
TEST RECORDING FRAMEWORK
Developing and testing forensic methods for digital audio requires a dataset that is acquired under certain basic criteria that allows it to be diverse and reproducible by others. For the tests detailed in this paper four different devices were used to make the test recordings that were ultimately uploaded and then downloaded to YouTube to analyze the resulting LTASS. These devices include the Tascam-DR07, Olympus WS-100, Olympus WS-853, and the Sony ICD PX-333. From these devices a list of all the recording settings were made and recordings were made using every possible combination of all the settings on each device. Some of the settings used include VOR, VSYNC, Low Cut Filter, mic sensitivity level, and a variety of preprogrammed recording settings such as lecture, interview, meeting, and voice notes settings (Table 1). To control variability each combination of settings was recorded a total of three times. In total the dataset for this test consists of over 350 recordings. In each recording the make and model of each device, serial number, encoding algorithm, bitrate, mono/stereo setting, and all other applicable recording settings were listed.
8


Table 1 List of Device Settings
Recorder Name Device Setting
Socy ICD PX333 Low Cut Filter Mic Sensitivity (Low. Medium, High) VOR Lecture Meeting Voice Notes Interview 8kbps Mono Mp3 48kbps Mono Mp3 128kbps Mono Mp3 192kbps Mono Mp3
Olympus WS 853 Low Cut Filter VCVA Vsync (1,2,3, and 5 seconds) 8kbps Mono Mp3 64kbps Mono Mp3 128kbpsStereo Mp3
Olympus WS100 HQ, LP, SP Mic Sensitivity (Low, Medium, High) VCVA
9


Table 1 cont'd
Tascam DR07 16bitwav48KHz
16 bit wav 44.1KHz
24 bit wav 48KHz
24 bit wav 44.1KHz
32kbps MP3 48KHz
32kbps MP3 44.1KHz
64 kbps MP3 48KHz
64kbps MP3 44.1KHz
96kbps MP3 48KHz
96kbps MP3 44.1KHz
128kbps MP3 48KHz
128kbps MP3 44.1KHz
192 kbps MP3 48KHz
192 kbps MP3 44.1KHz
256kbps MP3 48KHz
256kbps MP3 44.1KHz
320kbps MP3 48KHz
320kbps MP3 44.1KHz
Uploading and Downloading From YouTube
Since YouTube does not allow for the upload of just audio with no video content all test recordings were combined with the same static video using FFmpeg as discussed earlier in Chapter 1 (Figure 2). The resulting files were then manually
10


uploaded to YouTube using the host platform. After all the videos had been uploaded Youtube-dl was used to query each file for the best quality DASH audio available (Figure 3). Youtbe-dl was then use to download the best quality audio available from each video using the command shown below.
% youtube-dl -f 140 -ict https://youtu.be/vid-ID-string In this example 140 would be replaced by the audio format code that represents the highest quality available from the query list. Before the test recordings were downloaded the YouTube account that was used to upload the videos was signed out to ensure the versions that were downloaded weren’t the original files, which are only accessible to the owner of the account used to upload them. Each of the downloaded files was then relabeled to identify them as the downloaded version.
MATLAB Analysis
In order to analyze the LTASS MATLAB was used to perform the analysis. MATLAB is multi-paradigm computing environment that is a commonly used tool for many forensic analyses. In this test the similarities between original exemplar recordings and their YouTube downloaded versions were compared to one another using Correlation Coefficient (CC) and Mean Quadratic Difference (MQD) of the LTASS and Power Spectral Density (PSD). The information gained from these plots show the level of compression that YouTube imparted on each of the test recordings after they were re encoding and downloaded using Youtube-dl.
11


CHAPTER IV
RESULTS
After analyzing the results from the LTASS plots and comparing their corresponding Correlation Coefficient (CC) values a few observations were made. When looking at the LTASS plots the most common loss in quality after recompression by YouTube was at and above the 15KHz range. This was not the case for all the plots but a majority of them showed these affects (see tables 4,6,9, and 10). In some cases the results between two recordings with different bit rate settings but otherwise identical device settings were more pronounced such as in Figure 4 and Figure 5 below which represent the plots of an 8kbps mono MP3 file and a 64kbps MP3 file from the same recorder (highlighted in Table 1). Similar results are also shown in Figure 6 and Figure 7 (highlighted in Table 1).
Another observation that was made is the relationship between the bit rate of the original recordings and their correlation to the CC values. The general trend across most of the recorders was the higher the bit rate of the original recording, lower CC values were found. This trend is clearly illustrated in Figure 15, which compares the bit rate and CC values of the Tascam DR-07 recorder and in Figure 11, which shows the bit rate and CC comparison for all recordings in the dataset. It was also observed that certain recorders showed more variability in their CC results and exhibited a lower average CC value, such as the Olympus WS853 as seen in Figure 16 which compares 8kbps recordings across all devices that record at that bit rate. This is also exhibited in Figure 12, which shows the Bit Rate vs. CC for the Olympus WS100.
12


LTASS: CC=0.95766, MQD=0.84707
Frequency (Hz) xiq4
Figure 4 LTASS Results for "OIympus_WS853_8kbps_MonoMP3_LCoff
LTASS: CC=0.99943, MQD=-0.057713
Frequency (Hz) x 1 q4
Figure 5 LTASS Results for "Olympus_WS853_64kbps_MonoMP3_LCoff'
13


Power/Frequency (dB/Hz)
LTASS: CC=0.98037, MQD=0.67365
Frequency (Hz) x 1 q4
Figure 6 LTASS Results for “Olympus_WS853_8kbps_MonoMP3_VSYNC_5sec_03"
LTASS: CC=0.9996, MQD=-0.10166
Frequency (Hz) x 1 q4
Figure 7 LTASS Results for “Olympus_WS853_64kbps_MonoMP3_VSYNC_5sec"
14


LTASS: CC=0.87695, MQD=1.2892
Frequency (Hz) xio4
Figure 8 LTASS Results for "Olympus_WS100_LP_MICSENS_high"
LTASS: CC=0.99648, MQD=0.55234
Frequency (Hz) x \ q4
Figure 9 LTASS Results for "Sony ICD PX333 48kbps Mono MP3 LCon MICSENS med"
15


LTASS: CC=0.9242, MQD=1.2893
Frequency (Hz) xiq4
Figure 10 LTASS Results For "Tascam DR0716bit Stereo Wav 44.1kHz"
Table 2 Outline of every file in the dataset along with the CC, MQD, sample rate before/after download, and the bitrate before/after download.
Test Recording File Name Correlation Coefficient Mean Quadratic Difference Original Bitrate Bitrate After Download Original Sample Rate Sample Rate After Download
Sony ICD PX333 8kbps MONO MP3 LCon MICSENS high 0.99864 0.17642 8kbps 96kbps 11.025KHz 44.1KHz
Sony ICD PX333 8kbps MONO MP3 LCon MICSENS low 0.99937 0.072384 8kbps 96kbps 11.025KHz 44.1KHz
Sony ICD PX333 8kbps MONO MP3 LCon MICSENS med 0.99932 0.1571 8kbps 96kbps 11.025KHZ 44.1KHz
Sony ICD PX333 8kbps MONO MP3 LCon VOR MICSENS high 0.99936 0.05124 8kbps 96kbps 11.025KHZ 44.1KHZ
Sony ICD PX333 8kbps MONO MP3 LCon VOR MICSENS low 02 0.99922 0.11038 8kbps 96kbps 11.025KHZ 44.1KHZ
Sony ICD PX333 8kbps MONO MP3 LCon VOR MICSENS med 0.99904 0.15531 8kbps 96kbps 11.025KHZ 44.1KHZ
16


Table 2 cont'd
Sony ICD PX333 8kbps MONO MP3 MICSENS high 0.99905 0.10662 8kbps 96kbps 11.025KHz 44.1KHZ
Sony ICD PX333 8kbps MONO MP3 MICSENS low 0.9995 0.1592 8kbps 96kbps 11.025KHZ 44.1KHZ
Sony ICD PX333 8kbps MONO MP3 MICSENS med 0.99948 0.1995 8kbps 96kbps 11.025KHZ 44.1KHZ
Sony ICD PX333 8kbps MONO MP3 VOR MICSENS high 0.99868 0.1865 8kbps 96kbps 11.025KHZ 44.1KHZ
Sony ICD PX333 8kbps MONO MP3 VOR MICSENS low 0.99858 0.28357 8kbps 96kbps 11.025KHZ 44.1KHZ
Sony ICD PX333 8kbps MONO MP3 VOR MICSENS med 0.99553 0.3844 8kbps 96kbps 11.025KHZ 44.1KHZ
Sony ICD PX333 48kbps Mono MP3 LCon MICSENS high 0.99095 0.62022 8kbps 96kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 48kbps Mono MP3 LCon MICSENS low 0.99944 0.0052299 48kbps 96kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 48kbps Mono MP3 LCon MICSENS med 0.99648 0.55234 48kbps 96kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 48kbps MONO MP3 LCon VOR MICSENS high 0.99437 0.55642 48kbps 96kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 48kbps MONO MP3 LCon VOR MICSENS low 0.99969 -0.094572 48kbps 96kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 48kbps MONO MP3 LCon VOR MICSENS med 0.99938 0.064902 48kbps 96kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 48kbps MONO MP3 MICSENS high 0.99955 0.0049564 48kbps 96kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 48kbps MONO MP3 MICSENS low 0.99908 0.015947 48kbps 96kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 48kbps MONO MP3 MICSENS med 0.99019 0.67009 48kbps 96kbps 44.1KHZ 44.1KHZ
17


Table 2 cont'd
Sony ICD PX333 48kbps MONO MP3 VOR MICSENS high 0.99943 0.59887 48kbps 96kbps 44.1KHz 44.1KHZ
Sony ICD PX333 48kbps Mono MP3 VOR MICSENS low 0.9992 0.077577 48kbps 96kbps 44.1KHz 44.1KHZ
Sony ICD PX333 48kbps MONO MP3 VOR MICSENS med 0.98432 0.78206 48kbps 96kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 Leon MICSENS high 0.99417 0.67404 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 LCon MICSENS low 0.98685 0.70185 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 LCon MICSENS med 0.99665 0.44112 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 LCon VOR MICSENS high 0.99854 0.43344 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 LCon VOR MICSENS low 0.98677 0.74819 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps Mono MP3 LCon VOR MICSENS med 0.9875 0.67995 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 MICSENS high 0.99444 0.53507 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 MICSENS low 0.98641 0.72433 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 MICSENS med 02 0.99147 0.67318 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 VOR MICSENS high 0.99536 0.54886 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 VOR MICSENS low 0.98707 0.75738 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 128kbps MONO MP3 VOR 0.99269 0.72581 128kbps 128kbps 44.1KHZ 44.1KHZ
18


Table 2 cont'd
MICSENS med
Sony ICD PX333 192kbps MONO MP3 LCon MICSENS high 0.96018 0.90092 192kbps 128kbps 44.1KHz 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 LCon MICSENS low 0.92841 1.1361 192kbps 128kbps 44.1KHz 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 LCon MICSENS med 0.95081 0.95192 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 LCon VOR MICSENS high 0.99037 0.82 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 LCon VOR MICSENS low 0.94887 1.1081 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 LCon VOR MICSENS med 0.96923 0.80867 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 MICSENS high 0.98345 0.69052 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 MICSENS low 0.94022 1.1037 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 MICSENS med 0.95941 0.90362 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 VOR MICSENS high 0.9887 0.81066 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 VOR MICSENS low 0.92653 1.1491 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 192kbps MONO MP3 VOR MICSENS med 0.94719 0.96923 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 Lecture 0.92848 1.0274 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 Meeting 0.95538 0.90807 192kbps 128kbps 44.1KHZ 44.1KHZ
Sony ICD PX333 VoiceNotes 0.98523 0.76471 128kbps 128kbps 44.1KHZ 44.1KHZ
Sony_ICD_PX333_lntervie w 0.97418 0.73401 192kbps 128kbps 44.1KHZ 44.1KHZ
Tascam DR07 16bit Wav 0.9242 1.2893 1536kbps 128kbps 44.1KHZ 44.1KHZ
19


Table 2 cont'd
44.1kHz
Tascam DR07 16bit Wav 48kHz 0.92193 1.2913 1536kbps 128kbps 44.1KHz 44.1KHZ
Tascam DR07 24bit Wav 44.1kHz 0.92523 1.2673 2304kbps 128kbps 44.1KHZ 44.1KHZ
Tascam DR07 24bit Wav 48kHz 0.91475 1.2681 2304kbps 128kbps 48KHz 44.1KHZ
Tascam DR07 32kbps MP3 44.1kHz 0.99877 0.21071 32kbps 128kbps 48KHz 44.1KHZ
Tascam DR07 32kbps MP3 48kHz 0.99279 0.60881 32kbps 128kbps 48KHz 44.1KHZ
Tascam DR07 64kbps MP3 44.1kHz 0.99696 0.4163 64kbps 128kbps 44.1KHZ 44.1KHZ
Tascam DR07 64kbps MP3 48kHz 0.97568 0.94223 64kbps 128kbps 48KHz 44.1KHZ
Tascam DR07 96kbps MP3 44.1kHz 0.99499 0.52883 96kbps 128kbps 44.1KHZ 44.1KHZ
Tascam DR07 96kbps MP3 48kHz 0.97591 0.88385 96kbps 128kbps 48KHz 44.1KHZ
Tascam DR07 128kbps MP3 44.1kHz 0.99677 0.41362 128kbps 128kbps 44.1KHZ 44.1KHZ
Tascam DR07 128kbps MP3 48kHz 0.9825 0.80648 128kbps 128kbps 48KHz 44.1KHZ
Tascam DR07 192kbps MP3 44.1kHz 0.99535 0.50164 192kbps 128kbps 44.1KHZ 44.1KHZ
Tascam DR07 192kbps MP3 48kHz 0.98009 0.83141 192kbps 128kbps 48KHz 44.1KHZ
Tascam DR07 256kbps MP3 44.1kHz 0.99814 0.31085 256kbps 128kbps 44.1KHZ 44.1KHZ
Tascam DR07 256kbps MP3 48kHz 0.97793 0.84058 256kbps 128kbps 48KHz 44.1KHZ
Tascam DR07 320kbps MP3 44.1kHz 0.99662 0.4307 320kbps 128kbps 44.1KHZ 44.1KHZ
Tascam DR07 320kbps MP3 48kHz 0.97987 0.92321 320kbps 128kbps 48KHz 44.1KHZ
Olympus WS100 HQ MICSENS high 0.99252 0.60786 32kbps 705.6kbps 44.1KHZ 44.1KHZ
Olympus WS100 HQ VCVA MICSENS Low 0.99908 0.17131 32kbps 705.6kbps 44.1KHZ 44.1KHZ
Olympus WS100 LP VCVA MICSENS high 0.90977 1.2368 5kbps 705.6kbps 8KHz 44.1KHZ
Olympus WS100 LP VCVA MICSENS Low 0.87338 1.289 5kbps 705.6kbps 8KHz 44.1KHZ
20


Table 2 cont'd
Olympus WS100 SP MICSENS high 0.84984 1.4207 16kbps 705.6kbps 22.05KHz 44.1KHZ
Olympus WSIOO SP MICSENS Low 0.86082 1.3502 16kbps 705.6kbps 22.05KHZ 44.1KHZ
Olympus WSIOO SP VCVA MICSENS high 0.86514 1.4098 16kbps 705.6kbps 22.05KHZ 44.1KHZ
Olympus WSIOO HQ VCVA MICSENS high 0.99916 0.4349 32kbps 705.6kbps 44.1KHZ 44.1KHZ
Olympus WSIOO LP MICSENS high 0.87695 1.2892 32kbps 705.6kbps 8KHz 44.1KHZ
Olympus WSIOO LP MICSENS Low 0.89148 1.179 5kbps 705.6kbps 8KHz 44.1KHZ
Olympus WS853 8kbps MonoMP3 Lcoff 0.95766 0.84707 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 Leon 0.99957 -0.11693 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 LCon VCVA 0.9801 0.73817 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 LCon VSYNC lsec 0.99839 0.27923 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 LCon VSYNC 2sec 0.99889 0.25384 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 LCon VSYNC 3sec 0.99828 0.33971 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 LCon VSYNC 5sec 0.99832 0.2308 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 VCVA 0.99923 0.27807 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 VSYNC lsec 0.99915 0.28521 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 VSYNC 2sec 0.99814 0.17792 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 VSYNC 3sec 0.99669 0.47638 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 8kbps MonoMP3 VSYNC 5sec 0.98037 0.67365 8kbps 96kbps 11.025KHZ 44.1KHZ
Olympus WS853 64kbps MonoMP3 LCoff 0.99914 0.028172 64kbps 96kbps 44.1KHZ 44.1KHZ
Olympus WS853 64kbps 0.9993 -0.012722 64kbps 96kbps 44.1KHZ 44.1KHZ
21


Table 2 cont'd
MonoMP3 LCon
Olympus WS853 64kbps MonoMP3 LCon VCVA 0.99772 0.28113 64kbps 96kbps 44.1KHz 44.1KHZ
Olympus WS853 64kbps MonoMP3 LCon VSYNC lsec 0.99959 -0.086612 64kbps 96kbps 44.1KHz 44.1KHZ
Olympus WS853 64kbps MonoMP3 LCon VSYNC 2sec 0.99946 -0.030395 64kbps 96kbps 44.1KHZ 44.1KHZ
Olympus WS853 64kbps MonoMP3 LCon VSYNC 3sec 0.99896 0.11884 64kbps 96kbps 44.1KHZ 44.1KHZ
Olympus WS853 64kbps MonoMP3 LCon VSYNC 5sec 0.99957 -0.091456 64kbps 96kbps 44.1KHZ 44.1KHZ
Olympus WS853 64kbps MonoMP3 VCVA 0.99573 0.34471 64kbps 96kbps 44.1KHZ 44.1KHZ
Olympus WS853 64kbps MonoMP3 VSYNC lsec 0.99888 0.13636 64kbps 96kbps 44.1KHZ 44.1KHZ
Olympus WS853 64kbps MonoMP3 VSYNC 2sec 0.99907 0.090133 64kbps 96kbps 44.1KHZ 44.1KHZ
Olympus WS853 64kbps MonoMP3 VSYNC 3sec 0.99951 -0.03995 64kbps 96kbps 44.1KHZ 44.1KHZ
Olympus WS853 64kbps MonoMP3 VSYNC 5sec 0.9996 -0.10166 64kbps 96kbps 44.1KHZ 44.1KHZ
Olympus WS853 128kbps StereoMP3 LCoff 0.98064 0.8661 128kbps 128kbps 44.1KHZ 44.1KHZ
Olympus WS853 128kbps StereoMP3 LCon 0.9819 0.84225 128kbps 128kbps 44.1KHZ 44.1KHZ
Olympus WS853 128kbps StereoMP3 LCon VCVA 0.98842 0.77031 128kbps 128kbps 44.1KHZ 44.1KHZ
Olympus WS853 128kbps StereoMP3 LCon VSYNC lsec 0.99152 0.74156 128kbps 128kbps 44.1KHZ 44.1KHZ
Olympus WS853 128kbps StereoMP3 LCon VSYNC 2sec 0.99344 0.67421 128kbps 128kbps 44.1KHZ 44.1KHZ
Olympus WS853 128kbps StereoMP3 LCon VSYNC 3sec 0.99256 0.67793 128kbps 128kbps 44.1KHZ 44.1KHZ
22


'PH:
.!*-! i
Bit rat vs. Correlation Coefficient for all recordings
1000 1500
Bit rate (kbps)
Figure 11 Plot of Bit Rate vs. Correlation Coefficient For All Recordings
Bit rate vs Correlation Coefficient for Olympus WSIOO
0.92
Bit rate (kbps)
Figure 12 Plot of Bit Rate vs. Correlation Coefficient For Olympus WSIOO
23


1 r
Bit rate vs. Correlation Coefficient for Olympus WS853
0.995
0.99
0.985
O
O
0.975
0.97
0.965
0.96
0.955
20
40
60 80 Bit rate (kbps)
100
120
140
Figure 13 Plot of Bit Rate vs. Correlation Coefficient For Olympus WS853
Bit rate vs. Correlation Coefficient for Sony ICD PX333
l
I
20 40 60
100 120 140
Bit rate (kbps)
160 180 200
Figure 14 Plot of Bit Rate vs. Correlation Coefficient For Sony ICD PX333
24


Bit rate vs. Correlation Coefficient for Tascam DR-07
1000 1500
Bit rate (kbps)
Figure 15 Plot of Bit Rate vs. Correlation Coefficient For Tascam DR-07
1 r
8kbps 44.1 KHz recordings and their corresponding CC Values
0.995
0.99 -
0.985
0.98
O
O
0.975
0.965
0.96
0.955 ----1
Average CC Values
Sony ICD PX333 0.998814167
Olympus WS853 0.992065833
Sony ICD PX333
13
Olympus WS853
Figure 16 Plot of 8kbps 44.1KFlz Recordings And Their Corresponding CC Values
25


1288kbps 44.1 KHz Recordings And Their Corresponding CC Values
1 r
0.998 -0.996 -0.994 -0.992 -q 0.99 -0.988 -0.986 -







• Tascam DR-07
• Sony ICD PX333
• Olympus WS853





0.984 -
0.982 -

0.98 L 0
Average CC Values
Tascam DR-07 0.99677
Sony ICD PX333 0.991493333 Olympus WS853 0.98808
j_____________i_________ 1 1
4 6 8 10 12
Recordings
Figure 17 Plot of 128kbps 44.1KHz Recordings And Their Corresponding CC Values
26


CHAPTER V
CONCLUSION
This thesis introduced and documented the collection of a general-purpose dataset of audio recordings from several recorders at all available settings. As a pilot study, this dataset was used in a study of YouTube effects on audio recompression. This was accomplished by analyzing the Long Term Average Sorted Spectrum (LTASS) of audio files before and after being uploaded to YouTube. Over 350 recordings were included in the dataset from a variety of recording devices and manufacturers to ensure a diverse dataset. All recordings were made under controlled conditions to ensure the results are easily comparable and reproducible.
In this study the most noteworthy finding was the relationship between the bit rate of the original recordings and their correlation to the CC values. The general trend across most of the recorders was the higher the bit rate of the original recording, lower CC values were found. This means that when higher bit rate recordings are uploaded and downloaded from YouTube using a tool such as Youtube-dl, more compression is detected compared to lower bit rate recordings. This study also revealed that the most common loss in quality after recompression by YouTube was at and above the 15KHz range. This was not the case for all the recordings but a majority of them showed these results.
Some possible areas of interest for future research could be: a deeper look into different audio compression analysis methods for audio that has been uploaded and then downloaded from YouTube to see if different results of what the re encoding process that YouTube automatically performs upon upload does to the audio files; a look into different source devices that record audio such as digital video cameras that record audio and videos
27


from cell phones to see how these files compare to those of handheld recorders after being re encoded by YouTube.
28


REFERENCES
[1] B. E. Koenig; D. S. Lacey, "Forensic Authentication of Digital Audio Recordings,"
Journal of the Audio Engineering Society, 09/2009, Volume 57, Issue 9
[2] N.C. Donnangelo, W.S. Kuklinski, R. Szabo, R.A. Coury, G.R. Hamshar, "Identification and Exploitation of Inadvertent Spectral Artifacts in Digital Audio," Journal of Digital Forensics, Security & Law. 2015; 10(3): 35-57.
[3] Whitecotton, C. M. (2017). "YouTube: Recompression effects," M.S., University of Colorado Denver. United States - Colorado, 2017
[4] "SWGDE." [Online]. Available:
https://www.swgde.org/documents/Current%20Documents/SWGDE%20Technical%20N otes%20on%20FFmpeg. [Accessed: 10-January-2019].
[5] Z. P. Giammarrusco, "Source identification of high definition videos: A forensic analysis of downloaders and YouTube video compression using a group of action cameras," M.S., University of Colorado at Denver, United States - Colorado, 2014.
[6] "ISO/IEC 23009-l:2014(en), Information technology— Dynamic adaptive streaming over HTTP (DASH)— Part 1: Media presentation description and segment formats." [Online]. Available: https://www.iso.Org/obp/ui/#iso:std:iso-iec:23009:-l:ed-2:vl:en. [Accessed: 21-February-2019].
[7] T. Painter, A. Spanias, "Perceptual Coding of Digital Audio," Department of Electrical Engineering Telecommunications Research Center Arizona State University. Proceedings of the IEEE ( Volume: 88 , Issue: 4 , April 2000 )
[8] Sinha, Deepen. "Audio Compression." AccessScience, McGraw-Hill Education, 2009.
[9] Anorga, ]., Arrizabalaga, S., Sedano, B. et al. "Analysis of YouTube’s traffic adaptation to dynamic environments" Multimed Tools Appl (2018) 77: 7977.
[10] Koenig BE, Lacey DS, Reimond CE. Selected Characteristics of MP3 Files Re-encoded With Audio Editing Software. Journal of Forensic Identification. 2014;64(3):304-321.
[11] "YouTube For Press," Google Developers. [Online]. Available: https://www.youtube.com/yt/about/press/. [Accessed: 12-Jan-2019].
[12] Video: Suspect Confesses Crime on YouTube. (2012, December 6). Local Broadcast Video Content. Available:
http://link.galegroup.com.aurarialibrary.idm.oclc.org/apps/doc/A311113718/AONE?u=a uraria_main&sid=AONE&xid=ebOf79bO. [Accessed: l-Feb-2019].
29


Full Text

PAGE 1

nrrrr rrr !"#$#%# &'# ()*+ ",!!!-.""'"#", &/-%"#$", &'-&"/,##%#$", !"#$#%# &'#0& "&%$-%$%%." #$", 1."!$# ",'2 #$ &!" #$// r/# '2 "! #2 &. ()*3

PAGE 2

4()*3 rrr rnrr

PAGE 3

,!",!!$# ",&!" #$//'2 ' 5r&r&# ,&!&00 #'$# ", r/# '2 "! #2 &. &"&% 2# &!,& 6$$.", #%,"/#""# &"7&*8()*3

PAGE 4

r&#' 5r&9r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

PAGE 5

nn nrr??????????????????????????? ??* -'##.0 !!#??????? ???????????????( .02?????????????????????????????= #--<'%??????????????????????????? @ #-!r!& /,??????????????????????? ?@ rrA????????????????????? ???+ &%&%# .&"!???????????????????? ???+ rrrrA?????????????? ?????8 0%#&'2&'#5%#&'2 #.# --*) &%!! ** r???????????????????????????? ???*( ????????????????????????? ???(B rr??????????????????????????????????? ?(3

PAGE 6

rn */!""2! /# ''$# &/, /# ' ????? ???????????????3 (-"%#$ $%",'&"&!"&%#25", ",C!&.0% &"$# D&$" '#5%#&'&'"," &"$# D&$" '#5%#&'?? ???????????????*+

PAGE 7

rnr r*E&.0%#$'2"&%,&',%' /# ' ????????? ??????????????( (.02/#..&'"#/#.E.0%& /# '25 ","!"'#???????= =E&.0%#$#-"-<'%1$# &&%&%$# .& "!#$&*3(F0!.##0=$%B @r!-%"!$# G%.0-!H8>=H8F0!H##=H #$$I????????*= >r!-%"!$# G%.0-!H8>=H+@F0!H##=H #$$?????????*= +r!-%"!$# G%.0-!H8>=H8F0!H##=H H>!/H)=I???*@ Br!-%"!$# G%.0-!H8>=H+@F0!H##= HH>!/I????*@ 8r!-%"!$# G%.0-!H*))HHH,2 ,I????????????*> 3r!-%"!$# G#J===@8F0!## =#.'I?*> *)r!-%"!# G&!/&.r)B*+"" #& @@*Fn;I?????????*+ **%#"#$"r&"!# %&"##$$/" # %%r/# '2!????????(= *(%#"#$"r&"!# %&"##$$/" # %.0-!*))???????(= *=%#"#$"r&"!# %&"##$$/" # %.0-!8>=???????(@ *@ %#"#$"r&"!# %&"##$$/"# # J===???????(@ *>%#"#$"r&"!# %&"##$$/" # &!/&.r<)B?????????(> *+%#"#$8F0!@@*An;r/# '2!', # !0#'2&%-!???(> *B%#"#$*(8F0!@@*An;r/# '2!', # !0#'2&%-!?(+

PAGE 8

rnrrn K#2 . &2# "'0/" -.K/"$/# F2 #-0#2"&%'/ =K<*-'#& <=K'#5!'&-'#K&$# .-'#%# .&"rK'&0""r&"nK&./'&0"" &.2# nnKn0 E" &!$ #"#/#%CK&C-&' &"/$$ /K# %&"##$$/"K-%!#'#'-%&"#K#2< .r/# '2.#'#%.0-!<*))nCKn2,C-&%" /# '2.#'#%.0-!<*))K"&'& ' /# '2.#'#%.0-!<*))#K#5-"%" ##$$K#5-"%" #$$K/!""!""2rK#/0 &"'r/# '2""2##J<= == K#//""2#%.0-!<8>=K& &%#" #%#//"-&"# !""2# %.0-!<*)

PAGE 9

E

PAGE 10

* r rnrn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
PAGE 11

( '#5%#&''$ #.#--&&%;2",'$$ /!"5",# 2&% /# '2! &'",-0%#&''D'#5%#&''$ #.#-!#! &/%& /#/0"#$,#5#-/#.0 !!!&-'#$ #.&& "#$'$$ "$%$ # .&"!&''/!""2!5%%2&' nr r ,%.&",0-%/.&#" &%;&-'#/#. 0 !!#!-!&%%& #-'-! '&$ #.",.-!/5%!""#5,&"5,& &'! #"#",!#-'!",&"& .""' $ #.&/,%'N!$&# ""#,&'"#$&-'#/ #.0 !!#2&",*38)N!5",", '"# '-/","< &" 1."$# #.0&/" !F9:5",#-"!&/ $/2 #"/&%'/ &!!&-'#1-&%"/", && "#$&-'#/#.0 !!# &%2# ",.!/#..#%F#5&!/#'/!,&" #'-/'",&"E0%#"",5&5,& "# '-/",!;#$$%!,.&5&",&"", !/#'/!'#",!!/&%%'0 /0"-&% /#'2LBM2",!0 #/!!",/#.0 !!#&% 2# ",.%##F!"#E0%#".0 $/"#!#$

PAGE 12

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rrrrr !" #-"-<'%!&$ / #!!<0%&"$# ./#..&'%"# #%",&"&%%#5!",-! "#'#5%#&' '#&'&-'#$ #.#--,-!#$",!"##% !/#!' '!"0 &/"//#.0& '

PAGE 13

@ "##", ", '0& "0 #2 &.!",&"00# ""#'#5% #&''#!$ #.#--&!"&%%#5! $# ",'#5%#&'#$&%%&&%&%$%"0!&' !#%-"#!$ #.#--N!! ! /#.0& '"##", "##%!5,/,#$$ &%."''#5 %#&'#0"#!# ","!"!'!/ ' ",!0&0 ",!"##%5&!-!'"#'#5%#&'",&-'# $ #.'#!",&"5 -0%#&''"# #--# ",00#!!#$",!"!"2.# 0 #'-/%",&//#-"",&"5&!-!' "#-0%#&'",'#!5&!#"-!'5,",'#!5 '#5%#&''-!2#-<'% #$%& "%/&-!#$",&"#$#--2& '#!" &.2! /#" ''/&"'"#&-'#",-%F#$ !& /, %&"2"# ",!"#0/,&!.0,&!;'", &-","/&"##$'#!",&",&
PAGE 14

> "/&."#'#5%#&'2'#!$ #.#--&&%; 2", !-%"2!" -/"&' ."&'&"&&$" &/,"##%5&!-!'"#'#5%#&'&' #$ #.#--,!&&%!!%&'"# ",-' !"&'2",&"",!"##%!&'", /#' 2",&"#--&-"#.&"/&%%'#! .&0-%&"'.-/,#$",."&'&"&$# .&"#", ,&' #$",$%5,/,,&!!" #2 .0%/&"#!$# $# !/&&%!!&'&-","/&" #00#!!L>M

PAGE 15

+ rr n'(n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n9n:'20%&&/F5,/,!&'&0" "r&"9r:'#!" &.2 "/,#%#2r&%%#5!#--"#!5"/,",'# &'&-'#1-&%"&!'#",-! ! &&%&%/#/"#O# 2&"&%E0%&!",&" G,.&#-"/#.#$",!$&"!",&" $# !"# 0%& 1-&%"0& &." !!"#P N !" /&&'&0""," &"#$ ",'#&!'#",/%"N!&&%&%'#5%#&' &'5'",!#",'#!" &.2! &'&0"'"#&'&./ #."IL3M #$"$ "$* # ' "#!",'$$ "&-'#1-&%"!",&" #--#$$ !'20%&&/F &!'#",-! !&&%&%&'5'",#-"-<'%/ &-!'"#1$# &%%&&%&% &-'#&''#$# .&"!2=%#5!,#5!& E&.0%#$",!1#$&&%&% $# .&"!#$&*3(F0!.##0=$%

PAGE 16

B $nr!%& r'r($

PAGE 17

8 rrr nr'(n) %#02&'"!"2$# !/.",#'!$# '2"&% &-'# 1!&'&"&!"",&"! &/1'-' / "&&!// " &",&"&%%#5! ""#' !&' 0 #'-/% #", !# ","!"!'"&%'",!0&0 $#'$$ "'/!5 -!'"#.&F", "!" /# '2!",&"5 -%".&"%-0%#&''&'" ,'#5%#&''"##--"#&&%; ", !-%"2,!'/!/%-'",&! /&.=&'",#J<=== #.",!'/ !&%!"#$&%%", /# '2!""2! 5 .&'&' /# '2!5 .&'-!2 0#!! %/#.&"##$&%%",!""2! #&/,'/#.#$",!""2!-!'/%-' r#5-"%" ./ !!""%%&'&& "#$0 0 #2 &..' /# '2!""2!!-/,&!%/"" 5."2&'#/#"!!""2!9&% *:#/#" #%& &%"&/, /#.&"##$!""2!5&! /# ''&"#"&%#$", ".!"#"&%",'&"&!"$# ",!"!" /#!!"!#$# =>) /# '2!&/, /# '2 ",.&F&'.#'%#$&/,'/ ! &%-. /#'2&%2# ",." &".##D!" #!""2&'&%%#", &00%/&% /# '2!""2!5 %!"'

PAGE 18

3 ")* + r/# ' &./""2 #/J===#5-"%" /!""9#5'-.n2,: r /""2 #/#"! " 5 8F0!##0= @8F0!##0= *(8F0!##0=*3(F0!##0= %.0-!8>=#5-"%" !/9*(=&'>!/#'!: 8F0!##0= +@F0!##0= *(8F0!" #0= %.0-!*))nC /!""9#5'-.n2,:

PAGE 19

"# *) &!/&.r)B*+"5&@8An; *+"5&@@*An; (@"5&@8An; (@"5&@@*An; =(F0!=@8An; =(F0!=@@*An; +@F0!=@8An; +@F0!=@@*An; 3+F0!=@8An; 3+F0!=@@*An; *(8F0!=@8An; *(8F0!=@@*An; *3(F0!=@8An; *3(F0!=@@*An; (>+F0!=@8An; (>+F0!=@@*An; =()F0!=@8An; =()F0!=@@*An; "$$+"$ /#--'#!#"&%%#5$# ",-0%#&'#$Q-! "&-'#5",#'#/#""&%% "!" /# '2!5 /#.'5",",!&.!"&"/ '#-!2.02&!'!/-!!' & % ,&0" *92(:, !-%"2$% !5 ",.&-&%%

PAGE 20

** -0%#&''"##---!2",,#!"0%&"$# .$" &%%",'#!,&'-0%#&'' #-"-<'%5&!-!'"#1&/,$%$# ",!" 1-&%"n&-'#&&%&%92=: #-"<'%5&!",-!"#'#5%#&'",!"1-&%" &-'#&&%&%$ #.&/,'#-!2 ",/#..&'!,#5%#5 nrn ",!E&.0%*@)5#-%' 0%&/'",&-'# $# .&"/#'",&" 0 !"!", ,2,!"1-&%"&&%&%$ #.",1%!"$ # ","!" /# '2!5 '#5%#&'' ",#--&//#-"",&"5&!-!'"#-0%#&'",' #!5&!!2'#-""#!", !#!",&"5 '#5%#&''5 N"",# 2&% $%!5,/,& #%&//!!%"#", #5 #$",&//#-"-!'"#-0%#&'",.&/,# $",'#5%#&''$%!5&!", %&%'"#'"$",.&!",'#5%#&'' !# '$", # ' "#&&%;",5&!-!'"#0 $# .",&&%!! !.-%"<0& &'2./#.0-"2 #."",&"!&/ #..#%-!'"##%$# .&$# !/ &&%!!",!"!"",!.%& "!"5# 2&%E.0%& /# '2!&'", #--'#5%#&'' !#!5 /#.0& '"##& #", -!2# %&"# #$$/"9:&'&C-&' &"/$$ /9C :#$",&'#5 0/" &% !"9:,$# .&"#2&'$ #.",!0 %#"!!,#5",%%#$/#.0 !!#",&" #--.0& "'#&/,#$","!" /# '2!&$" ",5 /#'2&' '#5%#&''-!2#-"-<'%

PAGE 21

*( r $" &&%;2", !-%"!$ #.",0%#"!& '/#.0& 2", /# !0#'2 # %&"##$$/"9:&%-!&$5#! &" #!5 .&',%##F2&"", 0%#"!",.#!"/#..#%#!!1-&%"&$" /#.0 !!##--5&!&"&' &#",*>An; &2,!5&!#"",/&!$# & %%",0%#"!-"&.&Q# "#$",. !,#5'",!&$$/"!9!"&%!@+3&'*): !#./&!!", !-%"!"5"5# /# '2!5",'$$ "" &"!""2!-"#", 5!'"/&%'/!""2!5 .# 0 ##-/'!-/,&!2@&'2> %#55,/, 0 !"",0%#"!#$& 8F0!.##=$%&'&+@F0!=$%$ #.", !&. /# ' 9,2,%2,"'&% *:.%& !-%"!& &%!#!,#52+&' 2B9,2,%2,"'&%*: #", #! &"#",&"5&!.&'!", %&"#!, 0"5"," &"#$", # 2&% /# '2!&'", /# %&"#"#", &%-!,2 &%" '&/ #!!.#!"#$ ", /# ' !5&!",,2, "," &"#$",# 2&% /# '2%#5 &%-!5 $#-',!" '!/%& %%%-!" &"'2*>5,/,/#.0& !"," &"&' &%-!#$",&!/&.r<)B /# ' &'2* *5,/,!,#5!"," &"&' /#.0& !#$# &%% /# '2!",'&"&!"" 5&!&%!##! '",&"/ "& /# ' ! !,#5'.# & &%"", !-%"!&'E, "'&%#5 & &2&%-!-/,&! ",%.0-!8>=&!!2*+5,/,/#.0& !8F0! /# '2!&/ #!!&%% '/!",&" /# '&"",&"" &",!!&%!# E,"'2*(5,/,!,#5!"," r&"!$# ",%.0-!*))

PAGE 22

*= ,)"-++./r01+23$020( (4$0)56 3)"-++./r01+23$07,0( (4$0)56

PAGE 23

*@ 7)"-++./r01+23$020( (4$08+!95030:$6 ; )"-++./r01+23$07,0((4$08+! 95036

PAGE 24

*> 2)"-++./r01+::0)40(<5+n9 +06 ')"-++./+<5*4=$$$,2( (4$)5(<5+n9+r6

PAGE 25

*+ :)"-++./"r*.:;7+ 1 ,,>?6 " 55@(A*@rB @B nr n r n r n !"###$%&!#' ()r)*+,,$-. *+/0-.1 $%&,-%&//+*12345 ..+/345 !"###$%&!#' (*+,,,#0 *+*01#$. $%&,-%&//+*12345 ..+/345 !"###$%&!#' (n*+,,,#1 *+/20/ $%&,-%&//+*12345 ..+/345 !"###$%&!#'6 ()r)*+,,,#*+*2/1. $%&,-%&//+*12345 ..+/345 !"###$%&!#'6 (*1*+,,,11 *+//*#$ $%&,-%&//+*12345 ..+/345 !"###$%&!#'6 (n*+,,,*. *+/22#/ $%&,-%&//+*12345 ..+/345

PAGE 26

"# *B !"###$%&!# ()r)*+,,,*2 *+/*--1 $%&,-%&//+*12345 ..+/345 !"###$%&!# (*+,,,2 *+/2,1 $%&,-%&//+*12345 ..+/345 !"###$%&!# (n*+,,,.$ *+/,,2 $%&,-%&//+*12345 ..+/345 !"###$%&!#6 ()r)*+,,$-$ *+/$-2 $%&,-%&//+*12345 ..+/345 !"###$%&!#6 (*+,,$2$ *+1$#20 $%&,-%&//+*12345 ..+/345 !"###$%&!#6 (n*+,,22# *+#$.. $%&,-%&//+*12345 ..+/345 !"###.$%&!#' ()r)*+,,*,2 *+-1*11 $%&,-%&..+/345..+/345 !"###.$%&!#' (*+,,,.. 7 *+**211,, .$%&,-%&..+/345..+/345 !"###.$%& !#' ( n*+,,-.$ *+221#. .$%&,-%&..+/345..+/345 !"###.$%&!#'6 ()r)*+,,.#0 *+22-.1 .$%&,-%&..+/345..+/345 !"###.$%&!#'6 (*+,,,-, 7*+*,.201 .$%&,-%&..+/345..+/345 !"###.$%&!#'6 (n*+,,,#$ *+*-.,*1 .$%&,-%&..+/345..+/345 !"###.$%&!# ()r)*+,,,22 *+**.,2-. .$%&,-%&..+/345..+/345 !"###.$%&!# (*+,,,*$ *+*/2,.0 .$%&,-%&..+/345..+/345 !"###.$%&!# (n*+,,*/, *+-0**, .$%&,-%&..+/345..+/345

PAGE 27

"# *8 !"###.$%&!#6 ()r)*+,,,.# *+2,$$0 .$%&,-%&..+/345..+/345 !"###.$%&!#6 (*+,,,1 *+*00200 .$%&,-%&..+/345..+/345 !"###.$%&!#6 (n*+,$.#1 *+0$1*.$%&,-%&..+/345..+/345 !"###/1$%&!#' ()r)*+,,./0 *+-0.*. /1$%&/1$%&..+/345..+/345 !"###/1$%&!#' (*+,$-$2 *+0*/$2 /1$%&/1$%&..+/345..+/345 !"###/1$%&!#' (n*+,,--2 *+..//1 /1$%&/1$%&..+/345..+/345 !"###/1$%&!#'6 ()r)*+,,$2. *+.##.. /1$%&/1$%&..+/345..+/345 !"###/1$%&!#'6 (*+,$-00 *+0.$/, /1$%&/1$%&..+/345..+/345 !"###/1$%&!#'6 (n*+,$02 *+-0,,2 /1$%&/1$%&..+/345..+/345 !"###/1$%&!# ()r)*+,,... *+2#2*0 /1$%&/1$%&..+/345..+/345 !"###/1$%&!# (*+,$-./ *+01.## /1$%&/1$%&..+/345..+/345 !"###/1$%&!# (n*1*+,,/.0 *+-0#/$ /1$%&/1$%&..+/345..+/345 !"###/1$%&!#6 ()r)*+,,2#*+2.$$/1$%&/1$%&..+/345..+/345 !"###/1$%&!#6 (*+,$0*0 *+020#$ /1$%&/1$%&..+/345..+/345 !"###/1$%&!#6*+,,1-, *+012$/ /1$%&/1$%&..+/345..+/345

PAGE 28

"# *3 (n !"###/,1%&!#' ()r)*+,-*/$ *+,**,1 /,1%&/1$%&..+/345..+/345 !"###/,1%&!#' (*+,1$./ /+/#-/ /,1%&/1$%&..+/345..+/345 !"###/,1%&!#' (n*+,2*$/ *+,2/,1 /,1%&/1$%&..+/345..+/345 !"###/,1%&!#'6 ()r)*+,,*#0 *+$1 /,1%&/1$%&..+/345..+/345 !"###/,1%&!#'6 (*+,.$$0 /+/*$/ /,1%&/1$%&..+/345..+/345 !"###/,1%&!#'6 (n*+,-,1# *+$*$-0 /,1%&/1$%&..+/345..+/345 !"###/,1%&!# ()r)*+,$#.2 *+-,*21 /,1%&/1$%&..+/345..+/345 !"###/,1%&!# (*+,.*11 /+/*#0 /,1%&/1$%&..+/345..+/345 !"###/,1%&!# (n*+,2,./ *+,*#-1 /,1%&/1$%&..+/345..+/345 !"###/,1%&!#6 ()r)*+,$$0 *+$/*-/,1%&/1$%&..+/345..+/345 !"###/,1%&!#6 (*+,1-2# /+/.,/ /,1%&/1$%&..+/345..+/345 !"###/,1%&!#6 (n*+,.0/, *+,-,1# /,1%&/1$%&..+/345..+/345 !"###'*+,1$.$ /+*10. /,1%&/1$%&..+/345..+/345 !"###r*+,22#$ *+,*$*0 /,1%&/1$%&..+/345..+/345 !"###6*+,$21# *+0-.0/ /1$%&/1$%&..+/345..+/345 8 8!"###8 9*+,0./$ *+0#.*/ /,1%&/1$%&..+/345..+/345 *0/-&:9*+,1.1 /+1$,# /2#-%& /1$%&..+/345..+/345

PAGE 29

"# () ..+/%45 *0/-&:9.$%45*+,1/,# /+1,/# /2#-%& /1$%&..+/345..+/345 *01.&:9..+/%45*+,121# /+1-0# 1#*.%& /1$%&..+/345..+/345 *01.&:9.$%45*+,/.02 /+1-$/ 1#*.%& /1$%&.$345..+/345 *0#1%&!#..+/%45*+,,$00 *+1/*0/ #1%&/1$%&.$345..+/345 *0#1%&!#.$%45*+,,10, *+-*$$/ #1%&/1$%&.$345..+/345 *0-.%&!#..+/%45*+,,-,*+./-# -.%&/1$%&..+/345..+/345 *0-.%&!#.$%45*+,02-$ *+,.11# -.%&/1$%&.$345..+/345 *0,-%&!#..+/%45*+,,.,, *+21$$# ,-%&/1$%&..+/345..+/345 *0,-%&!#.$%45*+,02,/ *+$$#$2 ,-%&/1$%&.$345..+/345 *0/1$%&!#..+/%45*+,,-00 *+./#-1 /1$%&/1$%&..+/345..+/345 *0/1$%&!#.$%45*+,$12 *+$*-.$ /1$%&/1$%&.$345..+/345 *0/,1%&!#..+/%45*+,,2#2 *+2*/-. /,1%&/1$%&..+/345..+/345 *0/,1%&!#.$%45*+,$**, *+$#/./ /,1%&/1$%&.$345..+/345 *012-%&!#..+/%45*+,,$/. *+#/*$2 12-%&/1$%&..+/345..+/345 *012-%&!#.$%45*+,00,# *+$.*2$ 12-%&/1$%&.$345..+/345 *0#1*%&!#..+/%45*+,,--1 *+.#*0 #1*%&/1$%&..+/345..+/345 *0#1*%&!#.$%45*+,0,$0 *+,1#1/ #1*%&/1$%&.$345..+/345 :/**4 ()r)*+,,121 *+-*0$#1%&0*2+-%&..+/345..+/345 :/**466 ('*+,,,*$ *+/0/#/ #1%&0*2+-%&..+/345..+/345 :/**'!66 ()r)*+,*,00 /+1#-$ 2%&0*2+-%&$345..+/345 :/**'!66 ('*+$0##$ /+1$, 2%&0*2+-%&$345..+/345

PAGE 30

"# (* :/**! ()r)*+$.,$. /+.1*0 /-%&0*2+-%&11+*2345..+/345 :/**! ('*+$-*$1 /+#2*1 /-%&0*2+-%&11+*2345..+/345 :/**!66 ()r)*+$-2/. /+.*,$ /-%&0*2+-%&11+*2345..+/345 :/**466 ()r)*+,,,/*+.#., #1%&0*2+-%&..+/345..+/345 :/**'! ()r)*+$0-,2 /+1$,1 #1%&0*2+-%&$345..+/345 :/**'! ('*+$,/.$ /+/0, 2%&0*2+-%&$345..+/345 :$2#$%& !#'*+,20-*+$.0*0 $%&,-%&//+*12345 ..+/345 :$2#$%&!#'*+,,,20 7*+//-,# $%&,-%&//+*12345 ..+/345 :$2#$%&!#'66*+,$*/ *+0#$/0 $%&,-%&//+*12345 ..+/345 :$2#$%&!#'6;/*+,,$#, *+10,1# $%&,-%&//+*12345 ..+/345 :$2#$%&!#'6;1*+,,$$, *+12#$. $%&,-%&//+*12345 ..+/345 :$2#$%&!#'6;#*+,,$1$ *+##,0/ $%&,-%&//+*12345 ..+/345 :$2#$%&!#'6;2*+,,$#1 *+1#*$ $%&,-%&//+*12345 ..+/345 :$2#$%&!#66*+,,,1# *+10$*0 $%&,-%&//+*12345 ..+/345 :$2#$%&!#6;/*+,,,/2 *+1$21/ $%&,-%&//+*12345 ..+/345 :$2#$%&!#6;1*+,,$/. *+/00,1 $%&,-%&//+*12345 ..+/345 :$2#$%&!#6;#*+,,--, *+.0-#$ $%&,-%&//+*12345 ..+/345 :$2#$%& !#6;2*+,$*#0 *+-0#-2 $%&,-%&//+*12345 ..+/345 :$2#-.%& !#'*+,,,/. *+*1$/01 -.%&,-%&..+/345..+/345 :$2#-.%&*+,,,# 7*+*/1011 -.%&,-%&..+/345..+/345

PAGE 31

"# (( !#' :$2#-.%&!#'66*+,,001 *+1$//# -.%&,-%&..+/345..+/345 :$2#-.%&!#'6;/*+,,,2, 7*+*$--/1 -.%&,-%&..+/345..+/345 :$2#-.%&!#'6;1*+,,,.7*+*#*#,2 -.%&,-%&..+/345..+/345 :$2#-.%&!#'6;#*+,,$,*+//$$. -.%&,-%&..+/345..+/345 :$2#-.%&!#'6;2*+,,,20 7*+*,/.2-.%&,-%&..+/345..+/345 :$2#-.%&!#66*+,,20# *+#..0/ -.%&,-%&..+/345..+/345 :$2#-.%&!#6;/*+,,$$$ *+/#-#-.%&,-%&..+/345..+/345 :$2#-.%&!#6;1*+,,,*0 *+*,*/## -.%&,-%&..+/345..+/345 :$2#-.%&!#6;#*+,,,2/ 7*+*#,,2 -.%&,-%&..+/345..+/345 :$2#-.%& !#6;2*+,,,7*+/*/--.%&,-%&..+/345..+/345 :$2#/1$%&!#'*+,$*-. *+$--/ /1$%&/1$%&..+/345..+/345 :$2#/1$%&!#'*+,$/, *+$.112 /1$%&/1$%&..+/345..+/345 :$2#/1$%&!#'66*+,$$.1 *+00*#/ /1$%&/1$%&..+/345..+/345 :$2#/1$%&!#'6;/*+,,/21 *+0./2/1$%&/1$%&..+/345..+/345 :$2#/1$%&!#'6;1*+,,#.. *+-0.1/ /1$%&/1$%&..+/345..+/345 :$2#/1$%&!#'6;#*+,,12*+-00,# /1$%&/1$%&..+/345..+/345

PAGE 32

(= 4C. 55 -. 4C. 55 r1+::

PAGE 33

(@ $4C. 55 r1+23$ ,4C. 55 +<5*4=$$$

PAGE 34

(> 34C. 55 "r*.%:; 742,,D>?.-" 5558

PAGE 35

(+ ;42,,D>?.-" 5558

PAGE 36

(B nrn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n; &2,!5&! #"",/&!$# &%%", /# '2!-"&.&Q# " #$",.!,#5'",! !-%"! #.0#!!%& &!#$" !"$# $-"!& /, /#-%'7&'0 %##F"# '$$ "&-'#/#.0 !!#&&%!!.",#'!$# &'#",&",&!-0%#&''&'", '#5%#&''$ #.#--"#!$'$$ " !-%"! #$5,&"", /#'20 #/!!",&" #--&-"#.&"/&%%0 $# .!-0#-0%#&''#!"# ",&-'#$%!R&%##F"#'$$ " !#/'/!",&" /# '&-'#!-/,&!'2"&% '#/&. &!",&" /# '&-'#&''#!

PAGE 37

(8 $ #./%%0,#!"#!,#5",!$%!/#.0& "# ",#!#$,&',%' /# ' !&$" 2 /#''#-

PAGE 38

(3 L*MA#2R&/G# !/-","/ &"##$2"&%-'#r/# '2!I 6#&%#$",-'#2 2#/")3D())3 #%-.>B!!-3 L(M#&2%#A-F%!Fr;&#r #rn&.!,& G'"$/&"#&' E0%#"&"##$&' ""0/" &% "$&/"! 2"&%-'#I E* @ +F) ()*>R*)9=:7=><>B L=M,"/#""#9()*B:G !"G.r@6(+@H 5 * H+I5@:;L@MGIL%M&%&%7 ,""0!7DD555!52'# 2D'#/-."!D"S()#/-."! DS()/,/&%S() #"!S()#S().02L//!!'7*)<6&-& <()*3M L>M&..& -!/#G#/'"$/&"##$ ,2,'$"#'#!7$# !/&&%!! #$'#5%#&' !&'#--'#/#.0 !!#-!2 &2 #-0#$&/"#/&. &!I !"#$#%# &'#&" "'"&"!<< #%# &'#()*@ L+MGD(=))3<*7()*@9:$# .&"#"/,#% #2T&./&'&0"!" &.2 # n9n:T& "*7'&0 !"&"#'!/ 0"#&'!2."$# .&"!I L%M&%&%7,""0!7DD555!## 2D#0D-DU! #7!"'7!#