Re: nitpicky about how smoothly speech synths can speed up
Hi all,
I'm also a non native english speaker, but I don't think the challenge would be to understand any concepts or new words in the text I would be reading, personally. To me, the constant speeding up annoys me. Sure, I use my voices at a medium to fast range, (my default voice is eSpeak for multilingual purposes, at rate 70 with boost off and my prefered reading voice for english on the iPhone is Victoria at rate 100 which I'm very comfortable with), and all these rates are about the highest I can go before everything starts falling apart for me. Eloquence used to be in my list for a few years, except now for whatever reason IBM TTS for NVDA kinda... Doesn't decide to work on my instalation for whatever reason, but I've come to accept eloquence as a lost synth and I'm perfectly fine with using Espeak. That's just my opinion though.
On a slightly related note, I agree with a point that has been stated earlier, where we were talking about how the voice needs to stay constant, or at least constant enough with it's inflections that you can focus on the content itself instead of the way the voice says things. I know a lot of people use Siri voices or vocalizer on their IOS devices, but me personally, I stick with Victoria precisely because of this fenomenon. Unit selection based concatonative voices, especially ones that are big, tend to have so much data in them for saying phrases that don't translate over for daily use, especially not with VO, for Siri itself, maybe, but not when reading quickly and not when you are trying to focus on what you're reading instead of the weird pronounciation glitches that can happen because of the way the units go together. This is really annoying when I'm reading in my native language, Hungarian, on IOS, since Mariska is the only voice available, and she, although only 43MB, has intonation and pronounciation errors in almost every sentence she says. And the annoying part is, I can't get used to the compact variants, either, at least not the vocalizer ones. Maybe the windows one core ones i can, but even then I find them a bit all over the place when it comes to inflection.
One thing that I'm not sure if anyone does, but I find the phonetic punctuation add on extremely useful for truncating bits of text or punctuation marks into tiny little beeps. The most common of these is a 2000hz tone for a comma, which make it so much faster to read through bits of text. With Espeak you can't tell if the rizing intonation at the end of a sentence is due to a comma, a question mark, or a semicolon, so I made the comma beep to differenciate it. Also, the elipsis has a sound of it's own to differenciate it as well. Besides the punctuation marks, very common things NVDA says, like when navigating object review, such as top, bottom, left, and right, are beeps as well. This also ties in to the audio themes add on, allowing me to know what kind of control I am focussed on in a moments notice instead of having to listen to it at the end of an utterance like I usually would with any other operating system. Things like this speed up my workflow enough that is comfortable for me, although if I have to use a temprary copy of NVDA on someone's computer, I can deel without them, too, of course, it's not necessary to have them, but when working regularly, at least on Windows with NVDA< it helps me, personally.
I hope this helped to bring more insight into the topic, and thanks for reading.
-- Audiogames-reflector mailing list Audiogames-reflector@sabahattin-gucukoglu.com https://sabahattin-gucukoglu.com/cgi-bin/mailman/listinfo/audiogames-reflector
