Making a Speech Recognition System that Understands Malayalam Words

Introduction Audio is the most effective means of communication used by humans. Automatic speech recognition can be defined as a technique that allows the system to recognize an input speech signal and interpret its meaning after which the system should be able to generate several control signals . 1 AIM The purpose of this project is to implement an automatic speech recognition system in hardware that understands a limited Malayalam word in the microphone. The system works well in an indoor environment (SNR is about 20 dB).

A speech recognition system limited to isolated words is called a discrete word system, and the user must pause between words. Experts mark speech recognition as one of the most difficult things about computers. Ultimately, the continuous word system will be able to interpret the continuous speech so that the user can talk normally; so far such a system has a single such as insurance or weather Have been limited by the subject vocabulary. The main advantage of providing input to a computer in normal conversation mode is ease of use. Such a system may also be caused by explosions of hand and wrist diseases associated with a wide range of computer keying. Today, the software makes it possible to listen to dictations from people who think that the computer is willing to pause. . . Speaking briefly. . Between. . . The best system is very accurate, equivalent to 70 words per minute.

Speech recognition is to accurately convert the waveform of our sound into actual words. In order for speech recognition to function effectively, it is important to understand the context of spoken words, such as knowing the meaning of "to" (or "two" or "too") based on surrounding words. Natural language processing is a process by which machines can interpret the grammar and context of human speech. Based on verbal intonation (equivalent to written punctuation), natural language processing is responsible for interpretation, such as how many people are being discussed in the phrase "Smith John and Sarah". This sentence can call John and Sarah both Smiths, Smiths, and two others (John and Sarah), depending on the grammar used.

Speech recognition (also known as speech recognition or text-to-speech conversion) converts speech to text. The machine uses microphones to capture our voice and transcribe our words in text. Using a simple text processing level, you can develop voice control functions using simple commands such as "turn left" or "call John". However, in order to achieve a higher level of understanding, a natural language understanding layer is required (see below). For example, the catering application - system MUST identify two distinct objectives: Food order - Order food - and Beverage order - OrderDrink. In this case, the NLU layer needs to understand that when the user says "please order food" it is OrderFood behind it. However, as people seldom speak like this, senior NLU needs to understand that the user's intent is also OrderFood when the user says "I'm hungry" or "I want to eat pizza" there is.