Speech Recognition Engines

Flat
HTK HCopy error when trying to convert .wav or .htk to .mfcc
User: Bjorndahl
Date: 12/18/2010 4:27 pm
Views: 16774
Rating: 14

Hello,

I'm trying to get myself trained on HTK with a baby data set.  Data was recorded at 44k, 16bit sampling rate on a Sony Devices 722 digital recorder in a soundproof booth in a .wav format.  These were processed first by PRAAT (as part of a different phonetic study), in case that makes a difference.

My problem arises when I try to use HCopy to convert the files to .mfcc.

I've read the forum post "converting .wav to .mfc using HCopy in HTK" (http://www.voxforge.org/home/forums/message-boards/speech-recognition-engines/converting-.wav-to-.mfc-using-hcopy-in-htk), but this doesn't solve my problem.

I had heard that there are problems using .wav with HTK, and so I converted my .wav files to .htk format using sox; either way I get the same problem. 

My analysis.conf file is as follows: (when I tried this with .wav files I changed SOURCEFORMAT to WAV).

***

# Acoustical analysis configuration file

#

SOURCEFORMAT = HTK # Format of speech files

TARGETKIND = MFCC_0_D_A # Identifier of the coefficients to use

#Unit = 0.1 micro-second

WINDOWSIZE = 250000.0 # = 25 ms = length of a time frame

TARGETRATE = 100000.0 # = 10 ms = frame periodicity

NUMCEPS = 12 # Number of MFCC coeffs (here from c1 to c12)

USEHAMMING = T # Use of Hamming function for windowing frames

PREEMCOEF = 0.97 # Pre-emphasis coefficient

NUMCHANS = 26 # Number of filterbank channels

CEPLIFTER = 22 # Length of cepstral filtering

ENORMALISE = T # Normalise intensity of data

***

I use the following command:

HCopy -A -D -T 1 C analysis/analysis.conf -S targetlist_htk.txt

but I get the following error
HTK Configuration Parameters[10]
  Module/Tool     Parameter                  Value
#                 ENORMALISE                  TRUE
#                 CEPLIFTER                     22
#                 NUMCHANS                      26
#                 PREEMCOEF               0.970000
#                 USEHAMMING                  TRUE
#                 NUMCEPS                       12
#                 TARGETRATE         100000.000000
#                 WINDOWSIZE         250000.000000
#                 TARGETKIND            MFCC_0_D_A
#                 SOURCEFORMAT                 HTK

  ERROR [+6310]  OpenParmChannel: cannot open Parm File /data/train/sig/Gr_F2_dhAsos_R1_aso.htk
  ERROR [+6313]  OpenAsChannel: OpenParmChannel failed
  ERROR [+6316]  OpenBuffer: OpenAsChannel failed
  ERROR [+1050]  OpenParmFile: Config parameters invalid
 FATAL ERROR - Terminating program HCopy

 

***

It seems that the more promising route is to go with the HTK files.  I tried the following:

HList data/train/sig/Gr_F2_dhAsos_R1_aso.wav 
But got:
ERROR [+1113]  IsWave: cannot read HTK Header in File data/train/sig/Gr_F2_dhAsos_R1_aso.wav
 FATAL ERROR - Terminating program HList

However, when I do the same with the .htk file, I get the following: (only head is pasted here, which is why it only shows 0 -- later samples have actual values)
----------------------------- Samples: 0->25717 -----------------------------
    0:       0      0      0      0      0      0      0      0      0      0
   10:       0      0      0      0      0      0      0      0      0      0
   20:       0      0      0      0      0      0      0      0      0      0
   30:       0      0      0      0      0      0      0      0      0      0
   40:       0      0      0      0      0      0      0      0      0      0
   50:       0      0      0      0      0      0      0      0      0      0
   60:       0      0      0      0      0      0      0      0      0      0
   70:       0      0      0      0      0      0      0      0      0      0
   80:       0      0      0      0      0      0      0      0      0      0

 

 

 

So it seems that my .htk files work, while my .wav ones are not readable by HTK at all.

I would appreciate any help possible.  I think I've provided all the necessary information, but in case it matters, I'm running HTK on Linux (OpenSUSE).

Thanks very much in advance for any help.

--- (Edited on 12/18/2010 4:27 pm [GMT-0600] by Bjorndahl) ---

Re: HTK HCopy error when trying to convert .wav or .htk to .mfcc
User: kmaclean
Date: 3/21/2011 12:23 am
Views: 292
Rating: 13

>TARGETKIND = MFCC_0_D_A

see this post: MFCC_D_N_Z_0 format

 

 

 

--- (Edited on 3/21/2011 1:23 am [GMT-0400] by kmaclean) ---

Re: HTK HCopy error when trying to convert .wav or .htk to .mfcc
User: ossia
Date: 6/28/2011 1:27 pm
Views: 162
Rating: 14

Hi 

I have the same exacrt problem while trying to get the .mfcc files. I however have been using HTK fromat ( .sig ) yet get the same series of errors you got. I was wondering if you could figure out what the problem was. I see you have posted this quite a while ago.

 

Thank you!

--- (Edited on 6/28/2011 1:27 pm [GMT-0500] by ossia) ---

Re: HTK HCopy error when trying to convert .wav or .htk to .mfcc
User: kmaclean
Date: 8/2/2011 8:44 am
Views: 2084
Rating: 15

cross posted here.

--- (Edited on 8/2/2011 9:44 am [GMT-0400] by kmaclean) ---

PreviousNext