Scott Guthery

I'm running 5prealpha downloaded and built (on a raspberry pi) about a week ago together with jsgf, en-us-phone-lm.bin and my own dictionary. I have silence SIL garbage +SPN+ in my dictionary and added these two words at various points in my jsgf constructs. pocketsphinx doesn't grump but neither of these words are recognized although [SPEECH] shows up in the backtrace listing at approximately the point that I inserted 'garbage' and <sil> shows up approximatly where I inserted 'silence'.</sil> What...

I.e. pocketsphinx_continuous doesn't do 6-grams, right? It does seem to handle 5-grams, however. At least it eats them and doesn't crash. Are you also signalling that pocketsphinx_continuous doesn't do <unk>?

Further to ARPA grammars: ERROR: "ngrams_raw.c", line 84: Format error; 6-gram ignored at line 25 INFO: lm_trie.c(474): Training quantizer Error in `pocketsphinx_continuous': double free or corruption (out): 0x0000000032cae150 \data\ ngram 1=1 ngram 2=1 ngram 3=1 ngram 4=1 ngram 5=1 ngram 6=1 \1-grams: -10.000 foo \2-grams: -10.0 foo foo \3-grams: -10.0 foo foo foo \4-grams: -10.0 foo foo foo foo \5-grams: 0.0000 when you hear <unk> \6-grams: 0.0000 when you hear <unk> \end\

I'll note in passing that lm_trie.c seems to enforce rules about the ARPA format that are over and above the format specifications. For example, if I have 0 1-, 2-, 3-, and 4-grams and 1 5-gram then lm_trie.c/recursive_insert fails the assertion priority_queue_size(ngrams)==0. If I just put the 5-gram in the lm file then ngram_model_trie.c/Line 458 fails with "Wrong magic header size number".

In the ARPA lm files generated by the CMU lmtool , , and <UNK> are not surrounded by apostrophes (') but in the examples of the ARPA format on the pocketsphinx help page regarding language models they are. Is there any differece in how they are treated?

My bad. Indeed it does. Strangely cross compiling pocketsphinx_continuous in the Windows/Eclipse/Artik IDE and then running on the Artik platform yields slightly different behavior (as judged by the Update from/Update to parameters) than compiling and running pocketsphinx_continuous on the Artik ... just enough different that the all the keywords were recognized by native pocketsphinx_continous but were NOT recognized by the cross-compiled pocketsphinx_continuous. It's probably a gcc thing. For now,...

As far as I can tell, ps_seg_iter on the C API does not return an iterator after a keyword search. Is there an analogous iterator for keyword searches that yields the same information as ps_seg_iter? Thanks in advance for any guidance. Cheers, Scott

It looks to me as if ad_read quits when gets to a buffer with WHDR_DONE set but doesn't bother to check if there are some (final) bytes in the buffer. Or?