Orca screen reader usage
I am a translator!
I tried orca in French since this is the langauge of my system and the language I am most comfortable with. The voice is not natural at all and so difficult to understand that it isn't really usable.
I tried changing settings, even with lower speed it remains very difficult to understand. I tried a number of voices, they are all equally difficult to understand.
Besides, when I try to change settings, the settings become largely unresponsive, making it difficult to adjust them.
Is anyone on this forum using orca successfully?
Is the problem only with French? Is English working fine?
I am a translator!
> Is the problem only with French? Is English working fine?
I just tried, I find English even more difficult to understand but I have more difficulties understanding English than French anyway, so hard to judge.
The default speed is 50, I manage to understand much better by setting the speed to 15 (both for French and English) but even so, the voice is very painful to listen to.
Orca's back-end is espeak by default. A newer back-end is espeak-ng . You can simply install it by apt-get (and set orca to use espeak-ng instead). A friend of mine - who is visually impaired - uses espeak with Persian language on his phone for his daily activities. He once told me it takes some time to get used to it. You should start with lower speeds. For example try this at terminal (-s option is speed):
espeak-ng -v fr-fr -s 100 "La France tire son nom des Francs (peuple germanique) qui en ont institué les premiers fondements sur les bases de la Gaule romaine" After you got used to it, you may increase the speed. Well-trained ears can listen to espeak at an incredible rate!
I am a translator!
Thanks. I installed espeak-ng and set it to be used by orca, it is better. The voice also sounds better by setting the pitch to 6.5 instead of the default of 5.0.
It is not a pleasant voice but it is more understandable.
I understand that using a screen reader requires training. I can use a computer visually only but it may not always be so then I'd rather get more familiar with screen readers and their configuration.
I've seen neural-network based speech synthesizer (Like Google's) that are perfectly pleasant to listen to. It is not surprising that neural-network models are much better. Unfortunately I don't know of any libre ones though :(
UPDATE: I was living under a rock to not see a lot of libre deep learning models are being developed! There are tons of them on GitHub, just search by "text-to-speech" tag; All are written in Jupyter notebooks. I asked help from a friend who is a python master to help me install them. I'll share the results with you if I succeeded.
UPDATE2: I contacted him. It is totally possible to use deep learning with only and only free software! Look at this video. Thanks for your great question Avron!
I am a member!
I've been using orca for years, and have the espeak English-US voice running at full speed. I've been using synthesizers, of varying quality, for years, so, may not be the best one to ask. I'd start off slow, as has already been suggested. Are you also trying to use a screen reader? I can see how this, and getting used to a small tts voice, can be challenging.