Academia.eduAcademia.edu

Figure 2 – uploaded by Abbas Ali

See full PDFdownloadDownload figure
Fig. 3. The architecture of end-to-end Arabic speech recognition. In the experiments 2-3 convolution and 4-5 recurrent layers have been tested.

Figure 3 The architecture of end-to-end Arabic speech recognition. In the experiments 2-3 convolution and 4-5 recurrent layers have been tested.

Related Figures (6)

LIST OF SPEECH CORPORA USED TO BUILD ACOUSTIC MODEL
LIST OF SPEECH CORPORA USED TO BUILD ACOUSTIC MODEL
Fig. 1. Arabic character-set with phonemes and description
Fig. 1. Arabic character-set with phonemes and description
LIST OF CORPORA USED TO TRAIN LANGUAGE MODEL
LIST OF CORPORA USED TO TRAIN LANGUAGE MODEL
COMPARISON OF DIFFERENT ARRANGEMENTS OF CONVOLUTIONAL AND RECURRENT LAYERS ALONG WITH NETWORK WIDTH AND RNN TYPE
COMPARISON OF DIFFERENT ARRANGEMENTS OF CONVOLUTIONAL AND RECURRENT LAYERS ALONG WITH NETWORK WIDTH AND RNN TYPE
COMPARISON OF THE PROPOSED APPROACH WITH EXISTING SYSTEMS TABLE IV
COMPARISON OF THE PROPOSED APPROACH WITH EXISTING SYSTEMS TABLE IV
Related topics:
EngineeringComputer ScienceRecurrent Neural NetworkBeam SearchWord error rate
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved