Recently done projects, using offline speech recognition, the whole long time, check a lot of methods, and finally completed, the online information a bit chaotic, and most of those people wrote, a group of people turn! Below I summarize, also for later people to do a convenient.
about the environment configuration I will not say, I just follow this tutorial step by step, is absolutely feasible. http://gaebolg.blog.163.com/blog/static/19826906820136232810723/
The following highlights the operation of the demo and the configuration of The Voice library ( Note: The above tutorial run demo last write a mess, you can start from the copy acoustic model and voice library to see my !) )
First, the demo run requires the following files:pocketsphinx/model/en_us under the LM folder ( . dic file, . lm files), and the hmm folder. If you feel trouble, you can download the files I provided. http://download.csdn.net/detail/u011747781/8034061
Here's how to set up your own Chinese voice library:
1. Write your own command set:
Create a new TXT file and enter the command, such as:
Baidu
Google
Click File, Save As, note to change the encoding to UTF-8 format! (just below the encoding option) is saved as a TXT file.
2. Using online tools http://www.speech.cs.cmu.edu/tools/lmtool.html write the TXT file to generate the file. Download tgz file can (there are all files), copy tgz compressed the LM file in the package, because only this useful (if it is English, dic file can be used directly, you do not have the method I described below to generate the dic file!! )
3. Create a new TXT file. In the Data folder I gave, there is a file called Standard.dic, which is the standard voice library, ctrl+f find the pronunciation you need to copy into your new TXT file.
The format is as follows:
Baidu B ai D u
After that, save as UTF-8 format and rename the. dic file.
4, this belongs to your voice library is built, just overwrite my Data folder corresponding DIC, LM file can!
Finally, the full demo http://download.csdn.net/detail/u011747781/8034195 is available on the Pro test
Android offline speech recognition (Pocketsphinx)