Academia.eduAcademia.edu

Figure 4 – uploaded by Hassan Satori

See full PDFdownloadDownload figure
In order to evaluate the performances of the application, we performed some experiments on different individuals each one of them was asked to utter 10 Arabic digits (as described before in sec. 3.1). We recorded the number of words that were correctly recognized, and then a mean recognition ratio for each tester was calculated (see table 4 and 5). Table 4: Test results for individual speakers, where M means Man and W woman.

Table 4 In order to evaluate the performances of the application, we performed some experiments on different individuals each one of them was asked to utter 10 Arabic digits (as described before in sec. 3.1). We recorded the number of words that were correctly recognized, and then a mean recognition ratio for each tester was calculated (see table 4 and 5). Table 4: Test results for individual speakers, where M means Man and W woman.

Related Figures (6)

Fig. 1: Sphinx-4 Architecture, the main blocks are the FrontEnd, the Decoder, and the Linguist.
Fig. 1: Sphinx-4 Architecture, the main blocks are the FrontEnd, the Decoder, and the Linguist.
Hence, the corpus consists of 5 repetitions of every digit produced by each speaker. Depending on this, the corpus consists of 300 tokens. During the recording session, each utterance was played back to ensure that the entire digit was included in the recorded signal. All the 300 (10 digits - 5 repetitions - 6 speakers) tokens were used for training and testing phases. Table.1 shows some of the recording system parameters.
Hence, the corpus consists of 5 repetitions of every digit produced by each speaker. Depending on this, the corpus consists of 300 tokens. During the recording session, each utterance was played back to ensure that the entire digit was included in the recorded signal. All the 300 (10 digits - 5 repetitions - 6 speakers) tokens were used for training and testing phases. Table.1 shows some of the recording system parameters.
Fig, 2: Waveforms for digits 4 (44:)!) speaker 2 trial 2. generated by the open source waveform editor wavesurfer [22].
Fig, 2: Waveforms for digits 4 (44:)!) speaker 2 trial 2. generated by the open source waveform editor wavesurfer [22].
Table 2: Phonemes symbols, used in the training of HMMs. collection of feature vectors. The circularity in this training process is resolved using the iterative Baum- Welch or forward-backward training algorithm [23].
Table 2: Phonemes symbols, used in the training of HMMs. collection of feature vectors. The circularity in this training process is resolved using the iterative Baum- Welch or forward-backward training algorithm [23].
Table 3: Excerpt from the dictionary of Arabic digits application. The following table shows an excerpt from the jictionary used for the training and recognition phases.
Table 3: Excerpt from the dictionary of Arabic digits application. The following table shows an excerpt from the jictionary used for the training and recognition phases.
Related topics:
Speech Recognition Arabic
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved