Adbot Progress Report 1

Download Report

Transcript Adbot Progress Report 1

Outline
I.
II.
III.
IV.
V.
Achievements
Project Outputs
Issues
Risk Analysis
Target for the next three months
Project Manager:
Aljon Rey P. Aniban
Members:
Emmanuel B. Cagadas
Anna Mae C. Yap
2:47 PM
1
Works that have been achieved over the past month
2:47 PM
2
Achievements (1)

Web Page Revision (2nd Semester):
 Revision of Project Objectives
o Utilization of the landmark-based audio fingerprinting
algorithm which was based on Dan Ellis’s version of the
algorithm used in Shazam’s music matching service.
o Conversion of the current MatLab simulation of the
algorithm to JAVA
 Revision of Methodology and Framework
o Segmentation of commercial audio template, extraction
of prominent points, incorporating these points into
pairs and efficient searching and matching of the stored
commercials to the audio stream input.
2:47 PM
3
Achievements (2)

Project Web
Page
2:47 PM
4
Achievements (3)

Matching of a single audio commercial to an audio of
mixed contents

Matching of multiple commercial queries to an audio of
mixed contents

Audio recording from microphone and audio playing in
JAVA
2:47 PM
5
progress milestones/deliverables achieved during the last month.
2:47 PM
6
Project Outputs
Revised Project Web Page
 Matching of multiple commercial queries
simulated in MatLab
 Recording of audio commercials via
microphone in JAVA
 Partial gathering of additional audio
commercial test data

2:47 PM
7
2:47 PM
8
Issues (1)
 Integration
of Octave and JAVA
 For ease of use of the functions present
in the MatLab simulation to JAVA by
forming the same structure in Octave.
 The need for boundary detection
 To lessen the computation load and memory
usage
2:47 PM
9
Issues (2)
Large file size of wav files compared to
other audio formats (i.e mp3)
 Probable increase in computation time
of the algorithm when Octave is
incorporated with JAVA due to outside
function calls.

2:47 PM
10
2:47 PM
11
Risk Analysis

Real time versus Offline Commercial
Detection
 The time and space complexity that will
be handled by the proposed system
given that it is in real time.
2:47 PM
12
2:47 PM
13
Targets for the Next 3 Months
Add boundary detection feature for
queried commercials. (Currently being
designed)
 Generate the equivalent Octave codes
from the simulated algorithm in MatLab
 Incorporate generated Octave codes
with the JAVA prototype of the
application.

2:47 PM
14
Targets for the Next 3 Months
Design the database prototype and data
gathering of additional commercials.
 Finalize the design of the whole
application (GUI, database and
underlying design of algorithm)
 Test the efficiency of the modified
algorithm.

2:47 PM
15
Thank you
2:47 PM
16