AAC | ADPCM | Convolution | DAFx | FFT | IIR | Mixer | MP3 | MPEG | MPEG-4

Hi,
Hello everyone.I have joined the group recently and Im
trying to develop an algorithm for the recognition of
environmental sounds form a random audio signal based
on a predefined set of environmental sounds.
The recognition algorithm follows the primary
step,which is the feature extraction(mel frequency
cepstral coefficients in my case) of the signal.It
basically involves the comparison of the features of
the audio signal to those of the signals in the
predefined set.I have been going through some of the
approaches such as dynamic time warping,linear vector
quantization,Hidden markov models,Perceptron Neural
Networks and Gaussian mixture models.Can anybody tell
me of any other approach applicable to this scenario.

Thanks,
Jagan
	

(You need to be a member of audiodsp -- send a blank email to audiodsp-subscribe@yahoogroups.com )

Hi,

perhaps you are interested in some papers about musical genre
classification, which seems to be a related task. There were e.g. a
number of publications at the last ISMIR conferences (for a complete
list of ISMIR papers, see: http://www.ismir.net/all-papers.html)

The selection of features and the used classifier depends on the
properties of your classes. For example, is a stationary feature set a
good description of the class, or do you have to take into account
typical variations of the feature set over time?

Kind regards,
(a)

Jagan Reddy wrote:
> Hi,
> Hello everyone.I have joined the group recently and Im
> trying to develop an algorithm for the recognition of
> environmental sounds form a random audio signal based
> on a predefined set of environmental sounds.
> The recognition algorithm follows the primary
> step,which is the feature extraction(mel frequency
> cepstral coefficients in my case) of the signal.It
> basically involves the comparison of the features of
> the audio signal to those of the signals in the
> predefined set.I have been going through some of the
> approaches such as dynamic time warping,linear vector
> quantization,Hidden markov models,Perceptron Neural
> Networks and Gaussian mixture models.Can anybody tell
> me of any other approach applicable to this scenario.
> 
> Thanks,
> Jagan
> 
> 
> 
-- 
dipl. ing.
alexander lerch

zplane.development
http://www.zplane.de
holsteinische str. 39-42
D-12161 berlin
 fon: +49.30.854 09 15.0
 fax: +49.30.854 09 15.5
	

(You need to be a member of audiodsp -- send a blank email to audiodsp-subscribe@yahoogroups.com )