[Bug 209311] Review Request: espeak - Software speech synthesizer (text-to-speech)

bugzilla at redhat.com bugzilla at redhat.com
Wed Nov 1 17:17:17 UTC 2006


Please do not reply directly to this email. All additional
comments should be made in the comments box of this bug report.

Summary: Review Request: espeak - Software speech synthesizer (text-to-speech)


https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=209311





------- Additional Comments From mtasaka at ioa.s.u-tokyo.ac.jp  2006-11-01 12:17 EST -------
Well, just a quick look at this package:

A. From http://fedoraproject.org/wiki/Packaging/Guidelines :
* Filesystem Layout (and others)
  - I think installing header files into /usr/include should be
    avoided, they should be installed under /usr/include/espeak.
  - By the way, why do you install only 'speak_lib.h'? I can find
    other header files.
  - http://espeak.sourceforge.net/index.html
-----------------------------------------------------
The project name speak had already been taken by 
another project on SourceForge (for a Windows TTS front-end) 
so I added a letter 'e' to the front to make eSpeak. For now, 
the program executable remains speak and is referred to 
as such in the documentation.
-----------------------------------------------------
    I think the binary name 'speak' is somewhat troublesome and
    recommend that the name should be 'espeak' (and the related
    documentation should be changed).
  - Check if /usr/share/espeak-data/soundicons/ is required as
    this is a empty directory. 

* BuildRequires:
  - By the way, is 'portaudio' requires for BuildRequires?

* Compiler flags
  - Fedora specific compiler flags are not passed and this leads
    to useless debuginfo rpm.

B. From http://fedoraproject.org/wiki/Packaging/ReviewGuidelines :
   = Perhaps nothing, however, I just did a quick look...

C. Other things I have noticed.
* Installation process.
  - Well, I think the installation process of this package is
    somewhat illegible. Especially, the need of explicit description
    of soname in spec file should be avoided as you have to check
    if soname of shared library is changed each time you build this.
    This should be something like:

--------------------------------------------------------
cd src
install -m 0755 libespeak.so.*.*.* %{buildroot}%{_libdir}
ln -sf libespeak.so.*.*.* %{buildroot}%{_libdir}/libespeak.so
/sbin/ldconfig -n %{buildroot}%{_libdir}/
--------------------------------------------------------

* Voice data
  - Well, I cannot verify the license issue of binary voice data
    named *_dict as this is a binary.
    If these data can be reproduced from ascii text files, it should
    do so. By a quick look, 'speak' has a option of '--compile=?'.

    Is it possible to recompile voice data *dict files by this?
    (Note: executing 'speak' binary requires 'portaudio' on BuildRequires.
     Also, by default 'speak --compile=?' needs superuser admin as
     it tries to access /usr/share/espeak so a patch is required to
     make 'speak' binary have the argument of output files)

-- 
Configure bugmail: https://bugzilla.redhat.com/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.




More information about the Fedora-package-review mailing list