Hello community, thank you in advance for the attention.
I'm trying to adapt the portuguese/brazilian acoustic model wich i downloaded in this link https://sourceforge.net/projects/cmusphinx/files/Acoustic%20and%20Language%20Models/
i created the corpus file with me sentences in portuguese, recorded it and everything was going weel until i reach the "collect statistics from the adaptation data."
When i run the following sequence of commands:
./bw \
-hmmdir br-pt \
-moddeffn br-pt/mdef.txt \
-ts2cbfn .ptm. \
-feat 1s_c_d_dd \
-svspec 0-12/13-25/26-38 \
-cmn current \
-agc none \
-dictfn br-pt.dic \
-ctlfn speech.fileids \
-lsnfn speech.transcription \
-accumdir .
the output i got is this:
FATAL: "mod_inv.c", line 358: Number of feature streams in mixture_weights file 1 differs from the configured value 3, check the command line options
Can you help me with anything?
Try https://github.com/alphacep/vosk-api, it supports Portuguese.
it seems that, that api has a poor community and documentation on the net, and i got into a big work starting the project to get to this point so i want to stick with cmu but thank for your advice
Last edit: Hugo Silva 2020-05-26
up!!
I trained my model but it has missing the phase-3 in module:20 during training and after training my hmm folder does not exist in model parameter folder.
Last edit: ijazulhassan 2020-07-27