Speech to text conversion for local ethnic languages of MP like Malavi, Bagheli etc
Abstract : In present industry, communication is the key element to progress. Passing on information, to the right person, and in the right manner is very important, not just on a corporate level, but also on a personal level. The world is moving towards digitization, so are the means of communication. Phone calls, emails, text messages etc. have become an integral part of message conveyance in this tech-savvy world. In order to serve the purpose of effective communication between two parties without hindrances, many applications have come to picture, which acts as a mediator and help in effectively carrying messages in form of text, or speech signals over miles of networks. Most of these applications find the use of functions such as articulatory and acoustic-based speech recognition, conversion from speech signals to text, and from text to synthetic speech signals, language translation amongst various others. In this review paper, we’ll be observing different techniques and algorithms that are applied to achieve the mentioned functionalities.
? It is the process of comparing the unknown pattern with existing sound reference pattern and computing similarity between them.
? We also built a script to handle Bengali spell checking so that only a single spelling instance of a word exists.
? We check whether there is any existing error in the input transcript. If any error is detected, the error stack which keeps the record of errors is printed.
? The basic idea is to compile and in- corporate knowledge from a variety of knowledge sources with the problem at hand.
? It is characterized by the use of machine learning methods. SMT is a data-driven approach which uses parallel aligned corpora and treats translation as a mathematical reasoning problem.
? The expansion of methodologies in the past decade and the introduction of new applications for automated translation have highlighted the limitations of adopting one single approach to the problems of translation.
• The proposed model requires the usage of the open sourced framework Sphinx 4 which is written in Java and provides the required procedural coding tools to develop an acoustic model for a custom language like Bengali.
• In this paper, we propose a technique that will convert Bengali speech into text in real time. Through microphone this system takes Bengali voice as input.
• Speech to text conversion is the process of dissecting discrete syllables or phonemes of recorded vocal audio and converting them to their literal transliteration .
? The size of the vocabulary plays an important role in determining the complexity, performance, and precision of the system.
? There are two basic structures in formant synthesis, parallel and cascade, but for better performance, some kind of combination of these 2 structures is used.
? Hybrid machine translation is widely used due to its inculcation of advantages of both rule- based as well as statistical machine translation techniques.
|