Voxin was: Re: Switching to Linux

[John G. Heim]

You don't really want to suggest that the quality of a synthesized voice 
is meaningless, do you?

First of all, confort isn't "fluff". Do you think it doesn't matter what 
the temperature of your office is? What if they took away your chair and 
made you stand all day?  If someone want to spend six buck for the sheer 
comfort they get from voxin it would make sense.

Secondly, I can understand more from boxin at a higher speech rate than 
I can with espeak. Most likely, when people are reporting that they find 
voxin easier to listen too, what they really mean is that they can 
understand it better. The reason they get tired listening to espeak is 
that they are straining to understand it. If you find espeak easier to 
understand, that's fine. But it's silly to dismiss the quality of the 
synthesized voice as meaningless.

If you're going to say the quality of the voice doesn't matter, you're 
going to go against just every review of every hardware speech 
synthesizer ever written. Everybody cares about the quality of the 
synthesized voice.





On 05/10/13 10:30, Mike Ray wrote:
>
> I listen to espeak for twelve or more hours a day and most of that time
> I'm concentrating on what I'm doing, not on how fluffy and cute the
> voice sounds.
>
> Mike
>
> On 10/05/2013 15:59, John G. Heim wrote:
>> As someone who uses voxin 8 to 10 hours a day, my opinion is that the
>> problems you mention below are minor compared to the clarity and
>> responsiveness of voxin.
>>
>> It only costs six bucks. If you have to listen to your workstation for
>> 8 to 10 hours a day, it's well worth it.
>>
>>
>>
>>
>>
>> n 05/09/13 17:57, Kyle wrote:
>>> According to Brandon McGinty-Carroll:
>>> # As I recall, voxen requires /dev/dsp or somesuch ancient sound API.
>>>
>>> As far as I know, this is correct, but it's a lot worse than that. Not
>>> only does Voxin require an ancient sound API, but it also requires
>>> ancient C libraries in order to function. The source code is either lost
>>> or is otherwise unavailable even to those who would maintain it, so it
>>> can't even be rebuilt against the latest C libraries or even get any of
>>> its numerous bugs fixed. It still crashes on words like c a e s u r e,
>>> which according to Google is a bitcoin client written in Python, and is
>>> also a rather common username on some non-blindness related forums. It
>>> also crashes on a rather common OCR error when recognizing the word
>>> Wednesday. I googled that one as well, and turns out it is a very common
>>> OCR scanning error, especially when scanning newspapers. I was
>>> especially seeing it in scanned newspaper archives from the late 1800's
>>> and early 1900's. There are also reports of random crashes that cause
>>> Voxin and other speech synthesis engines with the exact same codebase
>>> but different names to randomly kill the screen reader, and there is
>>> nothing anyone can do about it, because the source code is not available
>>> or is lost. Worse still is the fact that many companies are actually
>>> making a profit from licensing something so outdated, broken and
>>> unstable, but I guess that's no different from what Microsoft has been
>>> doing for years <smile>. It may fall on deaf ears for some reason, but
>>> my recommendation is to avoid Voxin and all the other voices like it.
>>> Use eSpeak, because it ships with most distros and just works. If you
>>> don't like the way eSpeak sounds, you can still get festival working,
>>> and Festival is capable of running some amazing free voices. There's
>>> also Pico, which is now supported natively in speech-dispatcher. All
>>> these voices sound better and work better than Voxin, which literally
>>> makes my head hurt.
>>> ~Kyle
>>> http://kyle.tk/
>>>
>>
>
>

-- 
---
John G. Heim, 608-263-4189, jheim at math.wisc.edu