i am working on Arabic Speech recognition , i trained some data to produce an
acoustic model
when training data ...
i faced an error when make triphone step started
the error i wased faced is about Callaoc
when i train 0.5 hour the training complete without building tree
thanks
callaoc (8991123)
my phonelist is about 1000 phone and the training data is about 10 hour
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
my phonelist is about 1000 phone and the training data is about 10 hour
You do not have enough memory to train a triphone model with 1000 phones.
Current clustering implementation is not very memory efficient and it requires
too much memory.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear nshmyrev
we tried to train based on syllaba , but we could not complete training
because of this error
FATAL_ERROR: "ckd_alloc.c", line 79: Calloc failed from itree.c(64)
upon your advice we tarin on ci
training went well, on decodinig we get
callaoc (8991123)
do you think if we increase ram we can overcome this issue
thank you
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
You need to change source code to allow memory-efficient training with big
phoneset. Right now it waste memory because the training process was not
designed for such a big phoneset.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
when we training in ci model there the training complete without error but
when
we start decoding : i faced the error
FATAL_ERROR: "ckd_alloc.c", line 79: Calloc failed from itree.c(64)
same error ..
thank you ,
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
we start decoding : i faced the error FATAL_ERROR: "ckd_alloc.c", line 79:
Calloc failed from itree.c(64) same error ..
Sorry, I'm not sure you provide precise information. The file itree.c is a
part of the trainer, not a part of the decoder. Anyway both decoder and
trainer do not support large phonesets. Maybe you'll succeed with pocketsphinx
decoder, but definitely not with sphinx3.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
hi all,
i am working on Arabic Speech recognition , i trained some data to produce an
acoustic model
when training data ...
i faced an error when make triphone step started
the error i wased faced is about Callaoc
when i train 0.5 hour the training complete without building tree
thanks
callaoc (8991123)
my phonelist is about 1000 phone and the training data is about 10 hour
You do not have enough memory to train a triphone model with 1000 phones.
Current clustering implementation is not very memory efficient and it requires
too much memory.
Dear nshmyrev
we tried to train based on syllaba , but we could not complete training
because of this error
FATAL_ERROR: "ckd_alloc.c", line 79: Calloc failed from itree.c(64)
upon your advice we tarin on ci
training went well, on decodinig we get
callaoc (8991123)
do you think if we increase ram we can overcome this issue
thank you
You need to change source code to allow memory-efficient training with big
phoneset. Right now it waste memory because the training process was not
designed for such a big phoneset.
when we training in ci model there the training complete without error but
when
we start decoding : i faced the error
FATAL_ERROR: "ckd_alloc.c", line 79: Calloc failed from itree.c(64)
same error ..
thank you ,
Sorry, I'm not sure you provide precise information. The file itree.c is a
part of the trainer, not a part of the decoder. Anyway both decoder and
trainer do not support large phonesets. Maybe you'll succeed with pocketsphinx
decoder, but definitely not with sphinx3.