DTWFF-pitch feature and faster neural network convergence for speech recognition
This paper presents the pre-processing of speech templates for artificial neural network (ANN). The processed features are pitch and Linear Predictive Coefficients (LPC) for input and reference templates, based on Dynamic Time Warping (DTW) algorithm. The first task is to extract pitch features usin...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English English |
| Published: |
Faculty of Electrical Engineering
2007
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/8061/ http://eprints.utm.my/8061/3/RubitaSudirman2007_DTWFFPitchFeatureandFasterNeural.pdf http://eprints.utm.my/8061/4/564200.pdf |
| _version_ | 1848891609823838208 |
|---|---|
| author | Sudirman, Rubita Salleh, Sh. Hussain Salleh, Shaharuddin |
| author_facet | Sudirman, Rubita Salleh, Sh. Hussain Salleh, Shaharuddin |
| author_sort | Sudirman, Rubita |
| building | UTeM Institutional Repository |
| collection | Online Access |
| description | This paper presents the pre-processing of speech templates for artificial neural network (ANN). The processed features are pitch and Linear Predictive Coefficients (LPC) for input and reference templates, based on Dynamic Time Warping (DTW) algorithm. The first task is to extract pitch features using Pitch Scale Harmonic Filter algorithm. Another task is to align the input frames (test set) to the reference template (training set) using DTW fixing frame (DTW-FF) algorithm. This is a time normalization process in which it is needed for data with unequal length. By doing time normalization, the test set and the training set are adjusted to the same number of frames. Having both pitch and LPC features fixed frames, speech recognition using neural network can be performed. A high recognition rate is obtained using combined features of DTW-FF and pitch for Malay digit words of 0-9, as high as 100% is achieved. Another task included in this paper is to find the optimal global minimum of the NN surface using the conjugate gradient algorithm to replace the steepest gradient descent in the back-propagation algorithm. Results showed that conjugate gradient algorithm is able to find a better optimal global minimum. |
| first_indexed | 2025-11-15T21:00:42Z |
| format | Article |
| id | utm-8061 |
| institution | Universiti Teknologi Malaysia |
| institution_category | Local University |
| language | English English |
| last_indexed | 2025-11-15T21:00:42Z |
| publishDate | 2007 |
| publisher | Faculty of Electrical Engineering |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | utm-80612013-11-28T03:19:08Z http://eprints.utm.my/8061/ DTWFF-pitch feature and faster neural network convergence for speech recognition Sudirman, Rubita Salleh, Sh. Hussain Salleh, Shaharuddin TK Electrical engineering. Electronics Nuclear engineering This paper presents the pre-processing of speech templates for artificial neural network (ANN). The processed features are pitch and Linear Predictive Coefficients (LPC) for input and reference templates, based on Dynamic Time Warping (DTW) algorithm. The first task is to extract pitch features using Pitch Scale Harmonic Filter algorithm. Another task is to align the input frames (test set) to the reference template (training set) using DTW fixing frame (DTW-FF) algorithm. This is a time normalization process in which it is needed for data with unequal length. By doing time normalization, the test set and the training set are adjusted to the same number of frames. Having both pitch and LPC features fixed frames, speech recognition using neural network can be performed. A high recognition rate is obtained using combined features of DTW-FF and pitch for Malay digit words of 0-9, as high as 100% is achieved. Another task included in this paper is to find the optimal global minimum of the NN surface using the conjugate gradient algorithm to replace the steepest gradient descent in the back-propagation algorithm. Results showed that conjugate gradient algorithm is able to find a better optimal global minimum. Faculty of Electrical Engineering 2007-06 Article PeerReviewed application/pdf en http://eprints.utm.my/8061/3/RubitaSudirman2007_DTWFFPitchFeatureandFasterNeural.pdf text/html en http://eprints.utm.my/8061/4/564200.pdf Sudirman, Rubita and Salleh, Sh. Hussain and Salleh, Shaharuddin (2007) DTWFF-pitch feature and faster neural network convergence for speech recognition. Elektrika, 9 (1). pp. 9-13. ISSN 0128-4428 |
| spellingShingle | TK Electrical engineering. Electronics Nuclear engineering Sudirman, Rubita Salleh, Sh. Hussain Salleh, Shaharuddin DTWFF-pitch feature and faster neural network convergence for speech recognition |
| title | DTWFF-pitch feature and faster neural network convergence for speech recognition
|
| title_full | DTWFF-pitch feature and faster neural network convergence for speech recognition
|
| title_fullStr | DTWFF-pitch feature and faster neural network convergence for speech recognition
|
| title_full_unstemmed | DTWFF-pitch feature and faster neural network convergence for speech recognition
|
| title_short | DTWFF-pitch feature and faster neural network convergence for speech recognition
|
| title_sort | dtwff-pitch feature and faster neural network convergence for speech recognition |
| topic | TK Electrical engineering. Electronics Nuclear engineering |
| url | http://eprints.utm.my/8061/ http://eprints.utm.my/8061/3/RubitaSudirman2007_DTWFFPitchFeatureandFasterNeural.pdf http://eprints.utm.my/8061/4/564200.pdf |