VoxForge
VoxForge
From this post:
Hi Ken. Thanks for pointing me towards that draft page. I'm running into the following error when I try to do the forced alignment with my own recording:
Error [+6510] LOpen: Unable to open label file downsampled.lab
Any idea what might be causing the error? It seems to work ok with the example files given on the draft page of audio segmenting.
Thanks,
Dan
--- (Edited on 8/23/2008 12:33 pm [GMT-0400] by kmaclean) ---
Hi DRF,
>Error [+6510] LOpen: Unable to open label file downsampled.lab
HTK is looking for the label file - which is the list of words in your text (in the order they appear in your text).
The label file is contained in your words.mlf file. Usually when creating acoustic models, you have many transcribed audio files. So rather than have many text label files to keep track of, you put them all in a single mlf file (mlf=multiple label file).
HTK assumes that the name of your label file is the same as your wav file - with the ".wav" suffix changed to ".lab". So make sure the prefix of your label file matches that of your audio file.
Ken
--- (Edited on 8/23/2008 12:33 pm [GMT-0400] by kmaclean) ---
Hi Ken,
About this question, I notice that htk3.3 doesn't need the list of .labs. But htk3.4 do need that. There is an issue about these .lab files. As you build a silence model, these files don't have "sp". There is going to be a warning about no training data about "sp", then the model updating stops.
--- (Edited on 3/7/2009 4:06 am [GMT-0600] by kalmanbucy) ---
Hi kalmanbucy,
Thanks for the feedback!
>There is an issue about these .lab files. As you build a silence model,
>these files don't have "sp". There is going to be a warning about no
>training data about "sp", then the model updating stops.
I think that the original post was with respect to forced alignment (creating word timestamps on a transcribed speech audio file) using a pre-existing acoustic model.
Your comment seems to be dealing with *training* an new acoustic model. So I am not sure they apply in this context.
Please clarify.
thanks,
Ken
--- (Edited on 3/7/2009 2:11 pm [GMT-0500] by kmaclean) ---
In the "A Tutorial Example" of HTKbook, we need to put .lab files with those .mfc in the feature folder as we create flat start monophones. Then we execute "HERest" three times to train the model. After that we add a model for "sp" in the step 7 to fix silence model, and use "HLEd" to generate a new trascription file for all phones plus "sp", called phones1, right. Then we need to train the models twice more. However, an issue comes up. The .lab files under .data/train/feature remains old, i.e. no "sp" was added to these .lab files while silence model was fixed.If we follow the htk tutorial to execute this below:
HERest -C config/config1 -I labels/phones1.mlf -t 250.0 150.0 1000.0 -S train.scp -H hmms/hmm5/macros -H hmms/hmm5/hmmdefs -M hmms/hmm6 lists/monophones1
A warning will pop up:
Pruning-On[250.0 150.0 1000.0]
WARNING [-2331] UpdateModels: sp[25] copied: only 0 egs
in HERestp:
It means model can't be updated due to little training data, because .lab will be searched in the process of training. Have you ever meet this trouble before?
--- (Edited on 3/9/2009 12:54 am [GMT-0500] by kalmanbucy) ---
reposted as a comment under Step 7 of the VoxForge Tutorial: Re: Error [+6510] LOpen: Unable to open label file downsampled.lab
--- (Edited on 3/15/2009 8:04 pm [GMT-0400] by kmaclean) ---