I have been trying to train an acoustic model for telgu but am getting a few errors. My model_parameteres folder and model_architecture folders are not being genersted although others are! It says that some words are not in my dictionary although they are. I have attached a screenshot of the training as well as my etc folder with the cfg file
Thanks Nickolay. Took your advice and was able to train the model.
But the decoding part is taking too long...is that normal.
Attaching the logdir for your reference
Hey Nickolay, is it okay to get the bleow erorrs in 2-3% of the files while training the acoustic model.
(ERROR: “backward.c”, line 430: Failed to align audio to transcript: final state of the search is not reached)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
ERROR: “backward.c”, line 430: Failed to align audio to transcript: final state of the search is not reached.
Sometimes audio in your database doesn’t match the transcription properly. For example the transcription file has the line “Hello world” but in audio actually “Hello hello world” is pronounced. The training process usually detects that and emits this message in the logs. If there are too many of such errors it most likely means you misconfigured something, e.g. you had a mismatch between audio and the text caused by transcription reordering. If there are few errors, you can ignore them. You might want to edit the transcription file to put the exact word which was pronounced. In the case above you need to edit the transcription file and put “Hello hello world” on the corresponding line. You might want to filter such prompts because they affect the quality of the acoustic model. In that case you need to enable the forced alignment stage during training. To do that edit the following line in sphinx_train.cfg:
$CFG_FORCEDALIGN = 'yes';
and run the training again. It will execute stages 10 and 11 and will filter your database.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Ya I tried that.....but it gave me an error saying that an align file was missing! and cant i ignore the errors since it is about 2 percent of my data.
My decoding is taking a really long time... I am running this on a VirtualBox though, will it affect the speed of the decoding significantly. I am attaching the link to the latest logdir and my files! https://drive.google.com/open?id=1WxcmcHZe-RqXVT5fuX7CmXlrhRFWYUJV
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Ya I tried that.....but it gave me an error saying that an align file was missing! and cant i ignore the errors since it is about 2 percent of my data.
My decoding is taking a really long time... I am running this on a VirtualBox though, will it affect the speed of the decoding significantly. I am attaching the link to the latest logdir and my files! https://drive.google.com/open?id=1WxcmcHZe-RqXVT5fuX7CmXlrhRFWYUJV
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
I have been trying to train an acoustic model for telgu but am getting a few errors. My model_parameteres folder and model_architecture folders are not being genersted although others are! It says that some words are not in my dictionary although they are. I have attached a screenshot of the training as well as my etc folder with the cfg file
https://drive.google.com/open?id=1wowwyjYp_TDYg3XuxH0iGs3DQaBkzdnP
https://drive.google.com/open?id=1Cv0ksW1WIb3-MQyLYJ8QN21pN1Mnb1Jf
You miss space after
<s>, this is what error message tells you.Overall refrain from pasting screenshots, otherwise you'll be an insect in your next life.
Second thing to learn is to trust the computers. If computer tells you you made a mistake you really made a mistake.
Last edit: Nickolay V. Shmyrev 2018-03-28
Thanks Nickolay. Took your advice and was able to train the model.
But the decoding part is taking too long...is that normal.
Attaching the logdir for your reference
Hey Nickolay, is it okay to get the bleow erorrs in 2-3% of the files while training the acoustic model.
(ERROR: “backward.c”, line 430: Failed to align audio to transcript: final state of the search is not reached)
https://cmusphinx.github.io/wiki/tutorialam/#troubleshooting
ERROR: “backward.c”, line 430: Failed to align audio to transcript: final state of the search is not reached.
Sometimes audio in your database doesn’t match the transcription properly. For example the transcription file has the line “Hello world” but in audio actually “Hello hello world” is pronounced. The training process usually detects that and emits this message in the logs. If there are too many of such errors it most likely means you misconfigured something, e.g. you had a mismatch between audio and the text caused by transcription reordering. If there are few errors, you can ignore them. You might want to edit the transcription file to put the exact word which was pronounced. In the case above you need to edit the transcription file and put “Hello hello world” on the corresponding line. You might want to filter such prompts because they affect the quality of the acoustic model. In that case you need to enable the forced alignment stage during training. To do that edit the following line in sphinx_train.cfg:
$CFG_FORCEDALIGN = 'yes';
and run the training again. It will execute stages 10 and 11 and will filter your database.
Last edit: Prithviraj 2018-04-05