Convert sphinx .arpa language model to .DMP file issue

General Discussion

User: Ian
Date: 9/13/2010 8:42 am

Views: 10036
Rating: 5

Hello everyone,

i have tried to convert sphinx .arpa language model to .DMP file with lm3g2dmp tool (more details see http://sphinx.subwiki.com/sphinx/index.php/Hello_World_Decoder_QuickStart_Guide)

but always get this error:

INFO: lm3g2dmp.c(766): 6966334 = #bigrams created

INFO: lm3g2dmp.c(767): 33180 = #prob2 entries

INFO: lm3g2dmp.c(774): 22420 = #bo_wt2 entries

INFO: lm3g2dmp.c(554): Reading trigrams

.FATAL_ERROR: "lm3g2dmp.c", line 633: Offset from tseg_base > 65535; reduce log2(bg_seg_sz)

the language model was created with vocabulary that have 37330 words, can anybody help me to fix this issue?

thanks in adcance!

Ian

--- (Edited on 9/13/2010 8:42 am [GMT-0500] by Ian) ---

Re: Convert sphinx .arpa language model to .DMP file issue

User: nsh
Date: 9/13/2010 9:16 am

Views: 4686
Rating: 4

> i have tried to convert sphinx .arpa language model to .DMP file with > lm3g2dmp tool (more details

1. Don't use outdated guides. You can find all required and up-to-date information on CMUSphinx on CMUSphinx website:

http://cmusphinx.sourceforge.net

2. Don't use lm3g2dmp, use sphinx_lm_convert instead

3. Don't ask CMUSphinx questions here. Before asking, search CMUSphinx forums, then ask question there.

--- (Edited on 9/13/2010 18:16 [GMT+0400] by nsh) ---

Previous • Next •


Username	Password