[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

catalog of ESpeak TTS server issues



"Robert D. Crawford" <rdc1x@comcast.net> writes:

> I tried the file.  Here is the output: 
*SNIP*
> tclespeak.cpp:596: error: invalid conversion from 'const char*' to 'char*'

Lukas's patch fixes the issue at line 596 of tclespeak.cpp.  I have
some additional problems on Grml, and those are fixed by the makefile.

I classified problems with the ESpeak TTS server into two categories:
build and usability. 

Build issues:
1. Newer versions of g++ and const (fixed by Lukas).
2. TCL 8.4 headers refer to the long long datatype, and that datatype is not
permitted in ISO C++ 98.  All makefiles that use TCL 8.4 need -Wno-long-long in
their CFLAGS.
3. Do distros disagree about the location of TCL 8.4 headers and libraries?
My TCL 8.4 headers are in /usr/include/tcl8.4, and the shared library is
/usr/lib/libtcl8.4.so

Here are some of the usability issues we've discussed.  Initials in parens
indicate the person who first described the issue.
RDC = Robert Crawford, CMB = Chris Brannon, and LL = Lukas Loehrer.

1. Character echo may be unusable.  (RDC).
2. Sometimes ESpeak fails to speak a line when moving point to that line (RDC).
3. Voices with richness = 0 are silent (RDC).
4. Occasional stutter when using word echo (CMB).
5. ESpeak can get confused about voice parameters.  (CMB)
6. Pauses around punctuation disturb the flow of speech.  Reading source code
is difficult (LL).

I'm not sure what causes 1, 2, and 4.
3, 5, and 6 are fixable.

Fix 3: ESpeak doesn't offer a richness parameter, so Emacspeak translates
richness to volume.
That is fine.  ESpeak allows volumes in the range 0 to 230, where 0
is absolute silence.  Any setting under 50 or 60 is practically inaudible,
IMHO.  When translating richness to volume, there should be
a minimum acceptable volume.  I would say that the minimum should be
approximately 65.

Fix 5: Send something other than SSML to ESpeak.  Perhaps DECTalk commands
could be interpreted by the server?
I'm willing to work on this.

Fix 6: Let the server change punctuation marks to words before synthesis.
The drawback is that we lose ESpeak's ability to pronounce punctuation in
a language-specific way.  The server needs punctuation tables for each
language that Emacspeak supports.  Currently, those are English and French.
I'm willing to work on this as well, but I need a list of the French
names for punctuation marks, since I don't know French.

-- Chris

-----------------------------------------------------------------------------
To unsubscribe from the emacspeak list or change your address on the
emacspeak list send mail to "emacspeak-request@cs.vassar.edu" with a
subject of "unsubscribe" or "help"



If you have questions about this archive or had problems using it, please send mail to:

priestdo@cs.vassar.edu No Soliciting!

Emacspeak List Archive | 2007 | 2006 | 2005 | 2004 | 2003 | 2002 | 2001 | 2000 | 1999 | 1998 | Pre 1998

Emacspeak Files | Emacspeak Blog | Search the archive