Hi All,
i implemented pocketsphinx in my android mobile app. Besides english, i use german language as well. After applying the german language package files i noticed my application size drastically increased beacause of one file.
I downloaded the language package from http://www.voxforge.org/de/Downloads The 'voxforge.lm.DMP' file size is 26MB. The same type of file for the english package('en-phone.dmp') is only 156.2kb.
Is there any smaller DMP file exist for german language?
How big should be this file normally?
What is the purpose of the DMP file?
Thanks for the answers in advance!
Alex
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The same type of file for the english package('en-phone.dmp') is only 156.2kb.
This model is for phonetic decoder, a totally different purpose. The model for German phonetic decoding does not exist yet, but there is English model for large vocabulary decoding which also is about 26mb in size.
Is there any smaller DMP file exist for german language?
I do not think you need a smaller model for large vocabulary, it might be too inaccurate. Do you need a large vocabulary decoding at all?
How big should be this file normally?
About 20-30Mb.
What is the purpose of the DMP file?
Language model describes probability of word sequences for decoding. That is which word preceeds which. This concept is explained in our tutorial
Hi All,
i implemented pocketsphinx in my android mobile app. Besides english, i use german language as well. After applying the german language package files i noticed my application size drastically increased beacause of one file.
I downloaded the language package from http://www.voxforge.org/de/Downloads The 'voxforge.lm.DMP' file size is 26MB. The same type of file for the english package('en-phone.dmp') is only 156.2kb.
Is there any smaller DMP file exist for german language?
How big should be this file normally?
What is the purpose of the DMP file?
Thanks for the answers in advance!
Alex
This language model for large vocabulary decoding
This model is for phonetic decoder, a totally different purpose. The model for German phonetic decoding does not exist yet, but there is English model for large vocabulary decoding which also is about 26mb in size.
I do not think you need a smaller model for large vocabulary, it might be too inaccurate. Do you need a large vocabulary decoding at all?
About 20-30Mb.
Language model describes probability of word sequences for decoding. That is which word preceeds which. This concept is explained in our tutorial
http://cmusphinx.sourceforge.net/wiki/tutorialconcepts