In this process, a text is composed of speech, speech-to-text conversion is performed.

In this mechanism, it involves two processes:
Natural Language Understanding
Natural Language Generation
a. Natural Language Understanding

We use natural language understanding to learn the meaning of given text. For NLU, we must understand nature and structure of each word.

i. Lexical Ambiguity
In this, words have multiple meanings
ii. Syntactic Ambiguity
Basically, in this syntactic ambiguity, the sentence having multiple parse trees.

iii. Semantic Ambiguity
Generally, in this, sentence having multiple meaning

iv. Anaphoric Ambiguity
Basically, in this phrase or word are presents. That is previously mentioned but has a different meaning.
b. Natural Language Generation

Basically, automatic text produced from structured data. That is in a readable format with meaningful phrases and sentences. Although, the problem of natural language generation is hard to deal with. It is a subset of NLP.

Natural language generation divided into three proposed stages:

i. Text Planning
Basically, it’s ordering of content in structure data.

ii. Sentence Planning
Generally, from structured data, we have to combine sentences to represent the flow of information.

iii. Realization
Basically, to represent text we use a grammatically correct sentence.