All geek questions in one place

What are the methods for recognizing words in a sound stream?

I'm very interested.

In a broad sense, how to do the following:

  • Word separation detection.
  • Syllables detection.
  • Compensate for normal vocabulary words.
+3
source share
1 answer

This is the subject of extensive research to this day. Usually, the construction of a model begins, based on a linguistic analysis of the language in which you will recognize and find all cases of separation of words and syllables. Then recognition is carried out mainly using hidden Markov models by signal.

Here are some links that may give you some of the best ideas:

http://lands.let.kun.nl/literature/eric.2004.2.pdf

http://www.asel.udel.edu/icslp/cdrom/vol4/778/a778.pdf

http://en.wikipedia.org/wiki/Speech_segmentation

+3

Source: https://habr.com/ru/post/1697613/

More articles:


All Articles