1. Exploding information
•Recent studies show
that most of the stored
data is in the form of
multimedia.
•Large volume of
multimedia data makes
it difficult to handle it
manually
•Need to have an 1 hr of TV broadcast across the world is 100 Petabyte.
automatic method to
Source: http://www.sims.berkeley.edu/research/projects/how-much-
organize and use it info/summary.html#tv
appropriately.
2. Audio indexing
Audio classification - An
Reason of choosing audio data
●
important step in building an
for study
audio indexing system
Easier to process
–
An audio indexing system
Contains significant information
–
Indexing – method of
●
organizing data for further
search and retrieval.
Example – book indexing
Audio Indexing – indexing
●
non-text data using audio
part of it
Source: J. Makhoul et. al. “Speech and language technologies for audio
indexing and retrieval”, in Proc. of the IEEE, 88(8), pp. 1338-1353, 2000.
3. Levels of information in audio signal
Subsegmental information
●
Related to excitation source characteristics
–
Segmental information
●
Related to system / physiological characteristics
–
Suprasegmental information
●
Related to behavioural characteristics of audio
–
4. Missing component in existing
approaches and it's importance
Features derived based on spectral analysis
●
Carry significant properties of audio data at segmental level
–
Miss information present at subsegmental, suprasegmental level
–
Perceptually significant information in linear prediction
●
(LP) residual of signal
Complimentary in nature to the spectral information
–
Suprasegmental information not being used in current systems
–
5. EXPLORING
SUPRASEGMENTAL FEATURES
USING
LP RESIDUAL
FOR
AUDIO CLIP CLASSIFICATION
B.Yegnanarayana
Anvita Bajpai
yegna@iiit.ac.in
anvita@mailcity.com
International Institute of
Applied Research Group Satyam
Information Technology
Computer Services Ltd.,
Hyderabad
Bangalore
6. Audio clip classification
Closed set problem
●
To classify a given audio clip in one of the following
●
predefined categories
Advertisement, Cartoon, Cricket, Football, News
–
Issues in audio clip classification
●
Feature extraction
–
Effective representation of data to capture all significant properties of audio for
●
the task
Robust under various conditions
●
Classification
–
Formulation of a distance measure and rule/models
●
Training a models for the task
–
Testing – actual classification task
–
Combining evidences from different systems
–
9. Suprasegmental information in LP
residual for audio clip classification
Autocorrelation samples of Hilbert envelope of LP residual for 5 audio classes
10. Statistics of autocorrelation sequence
Correction – here we have statistics of autocorrelation sequence peaks of HE (not LP residual)
12. Summary & Conclusions
Need to organize multimedia data because of its
●
large volume and need in real-life applications
Shown presence of audio-specific
●
suprasegmental information in LP residual
Need to explore methods to use the
●
suprasegmental information as an additional
evidence for the audio clip classification task