![]() ![]() We explored three models, which consist of Convolutional Neu-ral Networks, Recurrent Neural Networks, and Connectionist Temporal Classification, towards end-to-end speech recognition on less-resourced language. The result can be integrated with other tasks such as spoken content retrieval. In this work, we focused on end-to-end speech recognition for less-resourced language, Amharic. Furthermore, recommendations have been provided for future researchers. The system outperforms an f-measure of 89.57%, 87.57%, 88.31%, 86.83%, 81.83%, and 87.59% for Amharic, Afan Oromo, Tigrinya, Hadiyyisa, Kambatissa, and Awngi languages respectively. Finally, precision, recall, and f-measures for each language have been computed following a successful evaluation of the model. Similarly, the user(s) can improve their skills in the selected languages accordingly. Here, the users can save time and perform their tasks easily. That means if the user(s) is (are) working on the Amharic language and then he/she/they can change the language she/he/they prefer(s) without shifting to another graphical user interface (GUI). ![]() In this model, the users can perform all spelling-related issues within a single system (all-in-one). The corpora used were gathered from a variety of sources, including economic, political, social, and related publications, newspapers, and magazines. In addition, the proposed model is evaluated using dictionary-based data sets for all languages. The major characteristics of the proposed model can be outlined by presenting suggestions for detected flaws and automatically correcting them utilizing the first suggestion. A dictionary-based methodology is used to detect and rectify various forms of misspelling-related issues. However, an effective and all-in-one typo detector and corrector system for Ethio-pian languages have yet to be developed. For some of these languages, there have been few works on typo detection and correction systems. In this paper, a misspelling detection and correction system was developed for Ethiopian languages (Amharic, Afan Oromo, Tigrinya, Hadiyyisa, Kambatissa, and Awngi). ![]() Such a term list, in addition to ease of compilation, has also an advantage in handling rare terms, proper nouns, and neologisms. Besides, instead of using a handcrafted lexicon for spelling error detection, we used a term list derived from frequently occurring terms in a text corpus. We get better result due to the smoothed language model, the generalized error model and the ability to take into account the context of misspellings. The proposed approach is evaluated with Amharic and English test data and has scored better performance result than the baseline systems: GNU Aspell and Hunspell. Since Amharic letters are syllabic, we used a modified version of the System for Ethiopic Representation in ASCII for transliteration in the like manner as most Amharic keyboard input methods do. The approach can be ported to other written languages with little effort as long as they are typed using a QWERTY keyboard with direct mappings between keystrokes and characters. It infers linguistic knowledge from a text corpus. We used a corpus-driven approach with the noisy channel for spelling correction. This paper describes an automatic spelling corrector for Amharic, the working language of the Federal Government of Ethiopia.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |