Ternyata ada hubungan antara audio dengan statsitika tepatnya matematika.
Seperti dalam mengklasifikasikan Audio,
Many audio analysis tasks become possible based on sound similarity and sound
classification approaches. These include:
• the segmentation of audio tracks into basic elements, such as speech, music,
sound or silence segments;
• the segmentation of speech tracks into segments with speakers of different
gender, age or identity;
• the identification of speakers and sound events (such as specific persons,
explosions, applause, or other important events);
• the classification of music into genres (such as rock, pop, classic, etc.);
• the classification of musical instruments into classes. Ternyata proses untuk menuju kesimpulan sama kayak analisis multivariate, Liat neh:
Analisa deskriptif -> segmentasi (analisis cluster) -> di pecah (analisis diksriminan) -> clasifikasi (analisis factor) ->decition deh
Wuihhh ternyata statistic kepake di dunia multi media….cihuyyyyyyyyyyyyyyy…
(berarti gw tetep setia dijalur gw...statistika dan multimedia)
Many classification systems can be partitioned into components such as :
1. A segmentation stage isolates relevant sound segments from the background
(i.e. the example violin sound from background noise or other sounds).
2. A feature extraction stage extracts properties of the sound that are useful for
classification (the feature vector, fingerprint). For both the sound similarity
and sound classification tasks, it is vital that the feature vectors used are
rich enough to describe the content of the sound sufficiently. The MPEG-7
standard sound classification tool relies on the audio spectrum projection
(ASP) feature vector for this purpose. Another well-established feature vector
is based on MFCC.
It is important that the feature vector is of a manageable size. In practice it is
often necessary to reduce the size of the feature vector. A dimension reduction
stage maps the feature vector onto another feature vector of lower dimension.
MPEG-7 employs singular value decomposition (SVD) or independent compo-
nent analysis (ICA) for this purpose. Other well-established techniques include
principal component analysis (PCA) and the discrete cosine transform (DCT).
3. A classifier uses the reduced dimension feature vector to assign the sound to
a category. The sound classifiers are often based on statistical models. Exam-
ples of such classifiers include Gaussian mixture models (GMMs), hidden
Markov models (HMMs), neural networks (NNs) and support vector machines
(SVMs).
Ternyata…statistika disini (audio integrator profesi gw neh) digunakan tapi dah userfriendly n saking friendlynya malah ga ketauan statistikanya…hmm sama aza ama para riset analisis pake SPSS yang mereka tinggal pake metode yang ada tanpa ia perlu tau bagaimana metode ini ada…hehehhehe sing penting tau data ini pake metode ini data itu pake metode itu…juga tujuan yg ingin dicapai….
Ternyata statistika berhubungan ma multimedia eh kebalik
multimedia -> statistika (baca: jika multimedia maka statistika) jadi kalo dibalik jadi pernyataan yang salah.(jadi inget mata kuliah PLMH)
Dimana :
