Abstract
For Improving the Business growth, the Business people try to know the customer's intension about their products. One of the best methods of collecting customer's feedback is telephone or mobile survey where customer service representatives(CSR) can interact with customers through phone calls and also record to analyze the customer's call data. The main issue of call data analysis through recorded files is a large amount of storage is required to store the audio files. This results increased costs, maintaining the hardware and software systems and manage a database system. In this paper we can directly convert the live calls into text files using speech to text (STT) algorithm and analyze these text files using Hadoop and MapReduce Framework for improving their future purchasing behavior.
Highlights
Speech [1] is the best way to communicate with others
There are some important methods to follow for converting speech to text that is Speech recognition, speech analysis, speech authentication
It is useful to read the output from console, with the help of file Writer object and print Writer class is used to print the output in text file
Summary
Speech [1] is the best way to communicate with others. It is nothing but expressing our feelings in vocalized form with others. It is used to identify a limited number of words and phrases with the help of machine or program or web applications in particular spoken language, converting speech to human readable format Figure 1. It works on acoustic and language modeling where Acoustic modeling [3] is used to represent the relationship between audio signals and phonemes (one unit of sound). Speaker dependent systems are mainly useful to recognize an particular user's voice those who are trained the system Linear predictive analyzer is best method for removing unnecessary data in speech In this method both features and parameters are involved. It is used to identify speakers, to adapt models like generate acoustic feature files, convert send-ump and mdef files, updating and recreating acoustic files, to align existing speech to text over timestamping
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have