GNOME Bugzilla – Bug 534932
flat review spelling broken since rev 3893
Last modified: 2008-05-27 16:53:40 UTC
Please describe the problem: in gedit, gnome-terminal, ff3, oowriter move flat review cursor, and try to get it to spell either the whole line, or just the current word. Currently it only does the first character and stops. rev 3892, the whole word, or sentance is spoken. This seems to have been carried forward into rev 3924. This is true for both military and normal spelling. Steps to reproduce: write: "this is a line." in gedit, put the flat review cursor on "this" and press orcakey+kk only "t" is said, it does not proceed to say "h i s". Actual results: Expected results: Does this happen every time? yes Other information:
Jon, I think this might be a duplicate of bug #532982. In particular see: http://bugzilla.gnome.org/show_bug.cgi?id=532982#c3 Are you using eSpeak, and do you have your speech rate set greater than (say) 75? If you slow it down, does it work? If you swap to another TTS (say Cepstral/Swift) and have the rate set high, does Orca spell the word correctly?
Created attachment 111560 [details] [review] does this help? That revision made two changes: 1. Dole out what needs to be spoken to the synth in a sayAll style 2. Interrupt the speaking of the line/word (which starts upon the first keypress) if we need to spell that line/word instead (which by default starts upon the second keypress). I'm wondering if the speech.stop() added for words is somehow kicking in too late. Since I can't repro this bug (even at a rate of 99), I can't verify that. The attached is just a guess. Jon, mind seeing if makes the problem go away for you? If it does, then it's probably worth checking in. worst case scenario would be that we finish speaking the word before spelling it, or finish spelling it before speaking it phonetically.
(In reply to comment #1) > Jon, I think this might be a duplicate of bug #532982. > In particular see: > http://bugzilla.gnome.org/show_bug.cgi?id=532982#c3 I am sorry, i searched for anything including "flat" in the bug table, but oversaw to check for "spell". It seems to be a duplicate. > Are you using eSpeak, and do you have your speech rate set > greater than (say) 75? If you slow it down, does it work? > If you swap to another TTS (say Cepstral/Swift) and have the > rate set high, does Orca spell the word correctly? using gnome-speech-espeak, all works fine, even when when set to very fast. When using speech-dispatcher and espeak (or any other voice), it fails to say anything but the first char.
(In reply to comment #2) > That revision made two changes: > 1. Dole out what needs to be spoken to the synth in a sayAll style > 2. Interrupt the speaking of the line/word (which starts upon the first > keypress) if we need to spell that line/word instead (which by default starts > upon the second keypress). > I'm wondering if the speech.stop() added for words is somehow kicking in too > late. Since I can't repro this bug (even at a rate of 99), I can't verify > that. The attached is just a guess. Jon, mind seeing if makes the problem go > away for you? If it does, then it's probably worth checking in. worst case > scenario would be that we finish speaking the word before spelling it, or > finish spelling it before speaking it phonetically. I'm afraid applying the patch seems to make no diffrence to me. speech routed via sd seems to only read first char, espeak and IBM viavoice using gnome-speech works fine (which seems to be the same before and after applying the patch).
Hi Jon, Joanie: > using gnome-speech-espeak, all works fine, even when when set to very fast. Interesting. I'm using eSpeak with gnome-speech, and I can reproduce the problem at any rate over about 75-80. > I'm afraid applying the patch seems to make no diffrence to me. Me neither.
> Interesting. I'm using eSpeak with gnome-speech, and I can reproduce the > problem at any rate over about 75-80. Huh. Well, then I guess you own this/these Rich. ;-) I can't repro it to save my life (mind you, I haven't tried setting up speech-dispatcher). > > I'm afraid applying the patch seems to make no diffrence to me. > > Me neither. That makes the callback approach to spelling suspect. But as to why it's a problem or what to do about it, I'm afraid I have no clue. :-(
> That makes the callback approach to spelling suspect. Agreed. ;-) I'm closing this one as a duplicate of bug #532982. Let's work up a fix there. *** This bug has been marked as a duplicate of 532982 ***