GNOME.org
 

Re: [orca-list] New Linux user, needing higher quality speech


As someone who did a lot of DSP programming back when I could see, if I
could find a linguistics expert who could tell me what there is about
the voice spectrum of a real human that makes it sound like a real
human, I could have a go at mucking about with the espeak voices.

But because I can't see the spectrums any more I am stumped.

It's all very personal, but I personally have no problem with any of the
espeak voices.  And there are some female voices.


On 18/05/2015 19:55, B.Henry wrote:

I totally agree, and have never used the default espeak voice for very
long, and have suggested that they change the default.
I like some of the espeak voice/variant combinations, and have made
voice files for myself and others so that some of these nvda/espeak
voices can be used easily on Linux as well as creating a few new voices
which include variant data in there voice files.
Even so I normally use the U.S. English male voice with no variant added
or other modification as my default voice on Linux.
Actually the phone version of voices usually if not always filters out
the high end as they assume that phone hardware is not able to reproduce
the higher frequencies.
I am almost certain I tested  eloquence's phone spin and this was true
there. If your hearing loss is most notable with higher frequencies
having more emphasis on the midrange may make it appear to have more
highs to you?
As for whether people would like espeak more if another voice was chosen
I can only assume that some would and some would not. Once in a while I
hear someone say that people would not mind espeak so much if only the
nvda default voice was used as default in Linux...lol, but again I can
only guess that would help some people, and hurt others.
I do always try and remind new users of any synth to experiment with
tone, and anything else available for that synth, e.g. inflection, and
to try all the available voices and or variants.
I wish those who package espeak would add a couple of female voices at
least, and a couple o  the klatts in ready to use voice files. To do so
for all languages could obviously grow the voices sub-folder of
espeak-data  a lot, but at least do so for English where espeak is best
in my opinion.
Again I think the bottom line is that voice preference is a matter of
choice, and in many cases need as some people just can not understand
one voice and can another, so even if one synth is well maintained and
much cleaner from a coder/packagers point of view it may be of little
value to some people who can comfortably use an old patched together
speech engine.
Those huge old decktalk external hardware synths made  the best synth
speech in their day by a mile in my opinion, and still are one of the
easier to understand options.
I've not heard a decktalk box in years, but if my memory is even 50%
correct they sure sound better than decktalk software speech, to me.

That "to me" is the key of course, and no matter what any of us like I
hope we can all agree that it is very important to have as many choices
available to Linux users both for use with speech-dispatcher and self
voicing aps, not to mention speakup.
          No matter how mujch better Linux may be than the alternatives
a user who can't understand  available voices, or gets headaches using
them is not going to stick around to find out.
B.H. Registerd Linux User 521886 On 18/05/15 02:47 AM, Tony Baechler wrote:

Sorry if this is a duplicate post.

I've purposely avoided this discussion because I don't want to get
into the
open source versus proprietary software wars, but apparently my opinion
seems to differ from the norm, so here it is for what it's worth.

On 5/17/2015 4:05 PM, Luke Yelavich wrote:

Its worth noting that NVDA out of the box uses an espeak variant, and
I myself don't find that particular variant palletable at all. I
disable all espeak variants on all systems where I use espeak, as the
default voice is more acceptable to my ears.

I wonder whether people would think differently of espeak if no
variant was used by default with NVDA.

I have used lots of speech synthesizers over the many years I've been
using
computers, going back to the Echo on the Apple II.  I've also used the
Doubletalk LT, the DECtalk Express, DECtalk Access32 (not on Linux),
ESpeak,
Eloquence and who knows what others I'm forgetting.  I must say that the
only one of those I can stand to listen to for many hours at a time is
the
DECtalk Express.  I have yet to find any software speech that I like.  I
find the software DECtalk to be muffled and hard to understand.  I can
tolerate ESpeak in short bursts, but it gives me a headache.  I should
say
here that I've used ESpeak primarily in Linux and was not aware that NVDA
ships a variant.  I like MBrola and could get used to it, but it isn't
very
responsive.  Festival is OK but not great and even more sluggish.  If
I must
use software speech, I use Eloquence for telephones.  Apparently there
are
two versions, normal and telephone with the difference being that
telephone
has better support for higher frequencies.  I don't think that variant is
available except on Windows.

Call it loyalty, being stubborn or whatever you want, but I'm still using
the 2.6.32 kernel because it was the last one with Speakup support for
hardware speech.  I would really like to switch to Orca for daily
tasks, but
I'm still in Windows because Orca apparently doesn't have serial support.
Similarly, I would like to switch to NVDA as my main screen reader in
Windows, but it definitely has no serial synthesizer support.  That's
really
weird because it supports serial Braille displays and it looks like it
wouldn't be hard to write a serial DECtalk driver.  I am not a Python
programmer though and I wouldn't have any idea what I'm doing.  Even
among
hardware synths, I have yet to find anything with the voice quality and
clarity of the DECtalk Express.  Note that versions of the firmware after
4.2CD are rubbish and not recommended.  I like the Audapter, but they
are no
longer being manufactured.  I can't stand the Accent line.  The Echo
was OK
as far as the sound of the voice, but very robotic.

To close, if there is a way to use my serial synth with NVDA and Orca, I
would like to know about it.  I would be happy to test any drivers if
anyone
wants to write one.  I have a slight hearing loss, so perhaps that's my
problem, but I generally have a hard time understanding most software
speech.
_______________________________________________
orca-list mailing list
orca-list gnome org
https://mail.gnome.org/mailman/listinfo/orca-list
Visit http://live.gnome.org/Orca for more information on Orca.
The manual is at
http://library.gnome.org/users/gnome-access-guide/nightly/ats-2.html
The FAQ is at http://live.gnome.org/Orca/FrequentlyAskedQuestions
Log bugs and feature requests at http://bugzilla.gnome.org
Find out how to help at http://live.gnome.org/Orca/HowCanIHelp


_______________________________________________
orca-list mailing list
orca-list gnome org
https://mail.gnome.org/mailman/listinfo/orca-list
Visit http://live.gnome.org/Orca for more information on Orca.
The manual is at
http://library.gnome.org/users/gnome-access-guide/nightly/ats-2.html
The FAQ is at http://live.gnome.org/Orca/FrequentlyAskedQuestions
Log bugs and feature requests at http://bugzilla.gnome.org
Find out how to help at http://live.gnome.org/Orca/HowCanIHelp



-- 
Michael A. Ray
Analyst/Programmer
Witley, Surrey, South-east UK

"In the beginning there was Debian, and Ubuntu was without form, and void"

Eyes-free Linux:
http://eyesfreelinux.ninja/
[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]