Phoneme level transcription using our own model

[Jisna Johnson]

I want to obtain phoneme level transcription using my own model.I completed training and obtained 7.1% WER. I used the command :

pocketsphinx_continuous -hmm ./ -lm bird.arpa -dict bird.dic -infile Crow1.wav|tee output

I got word level transcription. Then I used the command :

pocketsphinx_continuous -hmm ./ -lm bird.arpa -dict bird.dic -allphone birdphone.lm.bin -infile Crow1.wav|tee output

And got an error showing :

ERROR: "allphone_search.c", line 557: Phonetic LM does not have SIL phone in vocabulary

How can I solve this problem?
Please help.

[Nickolay V. Shmyrev]

build allphone lm birdphone.lm.bin with the SIL phone.