It can be concluded that
the selected features not
only reduced the time
and memory.
quick recognition rate of Neural
Networks
for
Arabic speech.
Acoustic
Modeling
in
5.
The single speaker Speech
Recognition.
training is easier than IEEE
Signal
Processing
double
in
the
multi- Magazine,
ISSN:
1053
speaker tests.
- 5888, DOI: 10.1109/
6.
The recognition rate MSP.2012.2205597,
Vol.
4- CONCLUSION
depends on the number 29, Issue. 6, Nov. 2012 (82
From the presented work, of patterns to be classified. ­ 97).
it can be concluded that:
[3]
Jeff
Dalton,
Atul
1.
The
sectionalizing 5- References
Deshmane, "An approach
concept gives a good [1]PNKulkarni, PCPandey to increasing machine
recognition rate of Arabic & D S Jangamashetti: intelligence",
IEEE
speech.
Multi-band
frequency Potentials, Artificial Neural
2.
The energy of the compression for improving Networks, pp. 33 - 36, April
speech signal is a good speech perception by 1991.
feature for classification of listeners with moderate [4] M. Hamed. "A quick
words.
sensorineural hearing loss, neural
network
for
3.
The LPC and RC J.SpeechCommunication, computer vision of gray
coefficients can be used Volume 54, Issue 3, images", Circuits, systems
successfully for Arabic March 2012 (341 - 350). and signal processing
speech recognition.
https://doi.org/10.1016/j.
journal, Vol. 16, No. 1, pp.
4.
INN as a processing specom.2011.09.005
41 - 58, 1997.
technique
presents
a [2] G Hinton et al: Deep
Research
37

---