[DECtalk] New synthesizer Gnuspeech

[brandon T]

Hello,

I hope this can be made into an NVDA addon please, I really like its
quality for some reason.

Thanks,

Brandon

On 10/22/15, Carlos Fernandez via Dectalk <dectalk at bluegrasspals.com> wrote:
> Actually, it wouldn't. The pronunciation at this stage is not sufficient
> to make it usable in any context, let alone a high-demand one like
> screenreading. I hope they update the system, and it generates English,
> all right, but the system is far from good.
> I am also worried about the constant mentioning of NeXT in the articles
> home page. I don't yet know if this is based off some code from NeXT,
> but the fact that they recommend actually buying a computer to run an
> operating system that still undergoes Y2K bugs and is closed-source, to
> boot, makes me wonder who is doing the work on this anyway. I would
> instead throw my support behind another sample-less speech engine with
> which we are all familiar, NVSpeechPlayer. I have been working (very
> minimally, though I intend to increase my efforts) on porting this to
> GNU/Linux and other platforms. I also want to add multilingual support
> to the engine, as it can already generate a pretty good English sound.
> I have seen what I would describe as a quantity before quality aspect to
> these projects, both by those who write them and those who collect them.
> While I view any speech synthesizer as a curiosity, I try to keep in
> mind the fact that the point of any engine is to be useful for some
> purpose. This usually means either naturalness of sound and fidelity to
> the true aspect of human speech or ability to function with little
> processor and memory overhead and scale to fast speeds. Those engines
> that are so old to sound extremely unnatural (take SAM from the early
> 1980s, for example) get a pass from me, as they clearly didn't have the
> capabilities that more modern hardware and operating systems make
> available to newer projects. Many other sub-par engines have been
> created and discarded throughout the years (my go-to example is
> festival, which remains incapable of almost any task given it).
> Collecting these attempts, especially when they are self-described as
> first starts, is entirely counterproductive to improving the options.
> I don't want to sound like I am prematurely consigning this project to
> the useless pile. The system may improve. If it does, it may be worth
> using in the future. At this stage, it is little of significance. I
> would recommend interested parties to work on the code, rather than
> attempting to use any program capable of generating some surrogate of
> speech in cases for which it is clearly inadequate.
>
> Carlos
> On 10/22/2015 17:50, Jake Gross via Dectalk wrote:
>> It would be nice if this was made into an nvda addon or a sapi5 voice.
>> Jake
>>
>> On 10/22/15, Miguel Villagomez via Dectalk <dectalk at bluegrasspals.com>
>> wrote:
>>> In my opinion, it sounds like a mix of Orpheus's U.S. and U.K. English
>>> accent.
>>>
>>> On Thursday, October 22, 2015, mattias via Dectalk <
>>> dectalk at bluegrasspals.com> wrote:
>>>
>>>> sounds the tts have bad english
>>>>
>>>> Den 2015-10-22 kl. 14:36, skrev Tony Baechler via Dectalk:
>>>>
>>>>> Hi all,
>>>>>
>>>>> I thought people here might be interested in reading about this
>>>>> synthesizer:
>>>>>
>>>>> http://www.gnu.org/software/gnuspeech/
>>>>>
>>>>> It's still very much in early development, but looks potentially
>>>>> interesting.  There are currently two versions, one for the Mac and
>>>>> one
>>>>> for
>>>>> Linux and Unixlike operating systems.  I don't have a Mac and I can't
>>>>> comment on Mac accessibility.  The Mac version comes with a graphical
>>>>> application which lets you change speech parameters.  I compiled the
>>>>> source
>>>>> on 64-bit Ubuntu 14.04.3 Linux if anyone wants the binaries and
>>>>> libraries.
>>>>> If you do, please write off list to the address in my signature.
>>>>>
>>>>> If you're on Windows or don't want to compile from source, you're not
>>>>> left out.  I'll shortly be uploading a demo in mp3 so you can hear it
>>>>> actually reading a large amount of text.  Yes, it should be able to
>>>>> sing
>>>>> from what I've gathered from the above site.  Here is the mp3 demo:
>>>>>
>>>>> http://classicradio.us/iso/gnuspeech_gpl.mp3
>>>>>
>>>>> Share and distribute the above demo as much as you want, but
>>>>> attribution
>>>>> would be appreciated.  It's written in C and the dictionaries are in
>>>>> plain
>>>>> text, so they should be modifyable without the graphical application.
>>>>> I
>>>>> would be interested in people's opinions on this.
>>>>>
>>>>>
>>>> _______________________________________________
>>>> Dectalk mailing list
>>>> Dectalk at bluegrasspals.com
>>>> http://bluegrasspals.com/mailman/listinfo/dectalk
>>>>
>>>
>>> --
>>> Sent from Gmail Mobile
>>>
>> _______________________________________________
>> Dectalk mailing list
>> Dectalk at bluegrasspals.com
>> http://bluegrasspals.com/mailman/listinfo/dectalk
>
> _______________________________________________
> Dectalk mailing list
> Dectalk at bluegrasspals.com
> http://bluegrasspals.com/mailman/listinfo/dectalk
>


-- 
It is simply sitting silently, witnessing the thoughts, passing before
you. Just witnessing, not interfering not even judging, because the
moment you judge you have lost the pure witness. The moment you say
"this is good, this is bad," you have already jumped onto the thought
process.
-- Osho