[mary-users] Problems with redstart wav file recordings - Urgent!
Robert Schuon
rschuon at shoutomatic.com
Tue Apr 22 19:29:29 CEST 2014
Hi, everyone!
We have run into a serious issue with the .wav files recorded
by the redstart program on version 5.1-SNAPSHOT.
Below is a header from one of the .wav files (002.wav), and you can
easily see the issue. The last four bytes of the wav (RIFF) format header
always read FE FF FF FF (or FF FF FF FE to us humans), which SHOULD be
the number of samples in the data to follow.
Instead, it reads 4Gb, and the Festival speech tools choke when the
files are attempted to be read.
Header (44 bytes):
52 49 46 46 22 00 00 00 57 41 56 45 66 6D 74 20
10 00 00 00 01 00 01 00 44 AC 00 00 88 58 01 00
02 00 10 00 64 61 74 61 FE FF FF FF - Data follows this as "samples"
This is a very big deal, because the Festival speech tools simply error
out when run, since they cannot read the length of the file. Any ideas?
I have a bunch of data I recorded already, but even rerecording is
useless if the size is not represented correctly. I looked at the first
5 .wav
files, and they all read the same (incorrect) value for the size.
For those that are not familiar with the format, here is a very good link:
https://ccrma.stanford.edu/courses/422/projects/WaveFormat/
And here is the command, and what happens:
/Applications/VCTK/speech_tools/bin/sig2fv -window_type hamming -factor
2.5 -otype est_binary -coefs melcep -melcep_order 12 -fbank_order 24
-shift 0.01 -preemph 0.97 -pm
/Users/shoutomatic/Documents/MaryTTS/myvoice2/pm/001.pm -o
/Users/shoutomatic/Documents/MaryTTS/myvoice2/mcep/001.mcep
/Users/shoutomatic/Documents/MaryTTS/myvoice2/wav/001.wav WALLOC: failed
to malloc -2 bytes
Thanks!
Bob S.
--
ShoutOmatic.com
"Stop Typing, Start Shouting!"
More information about the Mary-users
mailing list