[hts-users:00134] Re: about training files

[Kishore Prahallad <skishore@xxxxxxxxxx>]

Hi Heiga,

From the data available at  
http://kt-lab.ics.nitech.ac.jp/~tokuda/SPTK/EXAMPLE/data/data.tar.gz

I wanted to view the contents of  sample.pdf, and you suggested that I 
should use the following command.

swab +l sample.pdf | dmp +i | less

What it is displaying is "
0       1073746642
1       1062188517
2       1057343085
3       1054048949
4       1051151622
5       1048457869
6       1044413388
7       1041567425
"

1. I think these numbers (You suggested that the first 4 bytes in the 
pdf file indicating the dimension of fetaure vector ) are too high, pls 
clarify, whether the command I used and the output is correct or not?

2. If I am not, I wanted to know how to view the contents of sample.pdf, 
and data.mcep.

3. A similar problem exists while viewing data,mcep (*using swab and 
dmp). How to know the formant of these files?

Thanks.
Kishore


Heiga ZEN (Byung Ha CHUN) wrote:

> Hi,
>
> eem1lz wrote:
>
> > If you donot mind, I want to ask another question. The speech files I 
> > have got are provided by CMU. These file are recorded as *.wav 
> > format. If I want to use SPTK, these files need to be converted to 
> > *.raw format or other formats 
>
> without header.
>
> > I was doing the conversion one by one, but I realized that it is 
> > foolish to do the conversion in that way if we need a large mount of 
> > files. so is there any tool that can do it automatically for 
> > multi-files.
>
>
> I always use sox to remove RIFF header like as follows:
>
> for wav ./wav/*.wav
> do
>   raw=./raw/`basename $wav .wav`.raw
>   sox -c 1 -s -w -t wav -r 16000 $wav -c 1 -s -w -t wav -r 16000 $raw
> done
>
> Please visit http://sox.sourceforge.net/ or type "sox -h" for details 
> about sox.
>
> Regards,
>
> Heiga ZEN (Byung Ha CHUN)

--
Best Regards,
Kishore Prahallad
http://www.cs.cmu.edu/~skishore