| === meetingology` is now known as meetingology | ||
| b4n | TheMuso: hi! | 09:43 |
|---|---|---|
| b4n | I'm working on a new SPD output module (landegave will submit it in a moment), and the engine itself doesn't support punctuation speech | 09:43 |
| b4n | it works on Windows with NVDA (an open source screen reader), but that's because NVDA translates punctuation to actual words itself, before feeding it to its output modules | 09:44 |
| TheMuso | b4n: I know speech-dispatcher does have different punctuation options, but withouth having a closer look, I cannot remember whether it relies on the synth to deal with it, or whether it does some processing itself. | 09:45 |
| b4n | it relies on the synth to do it AFAICS | 09:46 |
| b4n | and e.g. espeak can do it just fine | 09:46 |
| b4n | the question is, would you accept a change that permits to spd to do the translation itself? | 09:47 |
| b4n | it would be fairly complex, because it's not a straightfoward task; but I cuold base my work upon NVDA's, which is GPL2+ | 09:47 |
| b4n | and theirs apparently work well | 09:48 |
| b4n | TheMuso: relevant NVDA sources include the pre-processig module (https://github.com/nvaccess/nvda/blob/master/source/characterProcessing.py#L601) and the pre-processing rules (e.g. for en: https://github.com/nvaccess/nvda/blob/master/source/locale/en/symbols.dic) | 09:50 |
| TheMuso | Yes, I would certainly accept that. We could go in a couple of directions with this actually. First, we allow the user to choose whether the synth does the work, or spd itself. Then, later on, we expand things such that spd knows if a synth can do it, and if it can, it will let the synth do it, otherwise spd will do it. | 09:50 |
| TheMuso | I've been thinking about this on and off or a while, but there are more pressing matters to deal with at the moment. | 09:51 |
| b4n | actually having spd doing it has some advantages, like having a consistent punctuation set on all synths | 09:52 |
| b4n | but ok, great, I'll start working on that than :) | 09:52 |
| TheMuso | Yeah, this is true. | 09:54 |
| TheMuso | b4n: BTW there is a speech-dispatcher channel here on freenode. | 10:01 |
| b4n | oh | 10:01 |
| TheMuso | #speech-dispatcher | 10:01 |
| b4n | I'll join there then, would make more sense indeed | 10:02 |
| TheMuso | Pretty quiet most of the time, but is used on occasion. | 10:02 |
| b4n | OK, I'm there now | 10:02 |
| TheMuso | Ok cool. I am idling in there pretty much all the time. | 10:03 |
| landegave | TheMuso: hello. This is Raphaël who requested you for Kali. I would like now to propose another tts support. I need a branch baratinoo to make a pull request | 10:05 |
| TheMuso | landegave: Are you saying you'd like a copy of speechd master on github to make a pull request against? | 10:07 |
| b4n | (FWIW, I made the changes, but for future technical reasons I'll let landegave submit the branch so he can add stuff easily during review) | 10:07 |
| TheMuso | landegave: Also as above, there is a speech-dispatcher channel on Freenode, #speech-dispatcher | 10:07 |
| TheMuso | Ok no problem. | 10:07 |
| b4n | TheMuso: no, the question is, against what to submit the PR | 10:07 |
| b4n | master? a specific branch that doesn't yet exist? | 10:07 |
| landegave | TheMuso: yes I just yet see. Thanks. | 10:07 |
| TheMuso | Oh, against master. | 10:07 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!