Red Hat Bugzilla – Bug 967229
espeak -w not always producing sane audio file
Last modified: 2013-08-23 10:25:32 EDT
Created attachment 753154 [details]
Description of problem:
When I use espeak with the -w file.wav option sometimes it doesn't produce a sane audio file, although it can speak the text correctly.
Version-Release number of selected component (if applicable):
espeak-1.47.11-1.el6.x86_64 (rebuilt from src.rpm on RHEL6.4)
Steps to Reproduce:
1. cat test.txt | espeak -v europe/bg -w test.wav
2. ls -l test.wav
-rw-rw-r--. 1 atodorov atodorov 880 25 май 22,19 test.wav
3. totem test.wav - nothing
$ cat test.txt | espeak -v europe/bg --stdout > test2.wav
$ ls -l test2.wav
-rw-rw-r--. 1 atodorov atodorov 160982 25 май 22,20 test2.wav
$ totem test2.wav - es expected
Text is spoken correctly (in Bulgarian). The first audio file is garbage, it's too small and when opened with Totem it doesn't play anything. Length is 0 sec.
The second audio file is correct and plays correctly in Totem.
I'm using the latest espeak RPM compiled from the src.rpm in Koji on RHEL6 because earlier versions in EPEL doesn't support Bulgarian language. However this shouldn't affect ability to write audio file.
Created attachment 753155 [details]
Created attachment 753156 [details]
Created attachment 785736 [details]
You can workaround it by:
$ espeak -ftest.txt -v europe/bg -w test.wav
I am pushing it into rawhide. I can roll-back it later in case upstream will come with different solution.