Method and system for correcting based on voice identification results

Abstract

The invention provides a method and a system for correcting based on voice identification results. The method comprises the following steps of on the premise that the accuracy of converting voice into a text cannot reach 100%, blocking the text of text information by combining boundary information with a predefined rule, using the boundary information as the basis, the quickly editing, modifying and deleting the text block before or after the boundary information in the voice identification results, and quickly inserting new text information in the boundary information. Compared with the traditional keyboard modifying method, the method for correcting the text based on the text-blocking voice identification results has the advantage that the number of operation times is reduced, and the interaction experience of a user and an intelligent terminal is greatly improved.

Description

technical field [0001] The invention relates to the technical field of speech recognition, in particular to a correction method and system based on speech recognition results. Background technique [0002] With the advent of the mobile Internet era, various mobile terminals represented by smart phones are increasingly occupying people's lives, and touch screens have also become the mainstream screen configuration of smart phones. However, whether it is the physical keyboard of the mobile phone or the virtual keyboard on the touch screen, text input on the mobile phone is always not as quick and convenient as PC keyboard input due to size limitations. Therefore, it has become a major trend to perform text input on mobile phones through the function of speech recognition. Applications such as voice input, voice texting, and voice search are increasingly appearing in smartphones. [0003] For the current common voice input applications, after these recognition errors occur in...

AI Technical Summary

Problems solved by technology

In other words, the second type of error is very easy to appear in speech recognition in the real environment

And as long as it occurs once, the extra operation brought to the user will offset a large part of the keyboard operation saved by using voice input

[0006] It can be seen that although the method of using speech recognition for text input is feasible, the accuracy of speech recognition can never reach 100% because people sometimes speak ambiguously, especially in noisy environments, the speech recognition results are extremely likely to be ambiguous. There are various bugs that cause users to have to bring up the keyboard again to make manual changes, which greatly reduces the convenience of voice input

Images