-----Original Message-----
From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Damien Garwood
Sent: October 3, 2018 8:48 AM
To: nvda@nvda.groups.io
Subject: Re: [nvda] How can I hel to improve synthesizers

Hi there,
There is actually more chance of being able to improve ESpeak, as it is open-source. But you'll need a lot of expertise on the C programming language, vocal properties and phonetics, and digital audio processing.
To be honest, I can't stand OneCore because of its extreme lagginess, both before speaking and its punctuation pauses. Unfortunately, SAPI also seems to have plunged down a mountain, except, ironically, with NVDA itself.
Cheers,
Damien.

On 03/10/2018 12:25 PM, Walmir Schultz via Groups.Io wrote:

Hi,

I recently upgraded to Windows 10 mainly to use the OneCore Voices.

But I am disappointed.

The OneCore USA voices are very good, but I am from Brazil and the
Portuguese brazillian voices are very bad.

The spelling, for example, is done with the english pronunciation of
the letters. Very weird and difficult to understand.

The intonation is far behind of Espeak and there so many words that
are pronounced the wrong way. Irritating.

So, I am forced to use eSpeak most of the time, even being tired of
all those unpleasant robotic voices and, every time I need to use the
English language, change to one of the American OneCore voices, that
sound better, have good intonation, almost no mispronunciation, even
lacking some characteristics I like in eSpeak, as the different
intonation of capital letters.

So, as I know eSpeak will probably continue to sound robotic, the
solution is to improve OneCore voices. Then, how can I do that? How to
suggest improvements and correct errors in brazillian vocabulary?

To be honest, I really do prefer eSpeak. It is far ahead of OneCore,
at least in brazillian Portuguese. If only it sounded like OneCore...