diff options
Diffstat (limited to 'kttsd/TODO')
| -rw-r--r-- | kttsd/TODO | 95 |
1 files changed, 95 insertions, 0 deletions
diff --git a/kttsd/TODO b/kttsd/TODO new file mode 100644 index 0000000..27ef1a3 --- /dev/null +++ b/kttsd/TODO @@ -0,0 +1,95 @@ +_ Filters: + _ XHTMLtoSSML stylesheet needs work to provide better mappings for most web pages. + _ Add KNewStuff capability for folks to upload/download filter configs. + _ There should be a default set of filters configured when user first runs KTTSMgr. + If user presses Default button, it should ask user whether to replace all existing + filters with the standard set. The standard set should probably include + the Qt to Plain Text and Fix Festival Unspeakable Characters filters, but I'm not + sure about the latter when desktop is not ISO 8859-1. + _ The SBD Filter configuration dialog can be shrunk smaller than the widgets on it. + This apparently due to a bug in the QLabel widget when it contains RichText. If + forced to plain text, the problem won't happen (I am told). However, the WARNING + in the dialog needs to be bolded. What to do? +_ Handbook: + _ Mention that "Root element is" and "DOCTYPE" are case sensitive. "html" is not + the same as "HTML". It should stay this way, BTW. Proper xhtml requires lowercase + tags. +_ Audio Plugins: + _ Audio Plugins are loaded for each sentence. Instead, reuse the plugins from a pool + of loaded plugins. + _ BUG:116542. Apply patch and resolve compile issue. + _ At least one user experiences a freeze-up using ALSA plugin. Speak a web page, + pause the page, cause a knotify to speak, kttsd freezes. This appears to be an + ALSA device contention problem? In this situation, there are two alsaplayer objects; + one in pause state, and one attempting to speak notify. + _ akodelib does not stop instantly when pause() or stop() is called. See bug:107135. + _ Why must Player constructor take a 3rd "args" argument, while PluginProc constructor + does not. They are essentially the same. ?? + _ Add option to automatically turn down music player volume while speaking (via dcop). +_ Festival Interactive plugin: + _ Need someone who can read czech to try integrating this into KTTS: + http://www.freebsoft.org/festival-czech + _ New Italian voice for Festival. http://www.pd.istc.cnr.it/FESTIVAL/home/default.htm +_ EPOS plugin: +_ Command plugin: + _ Command plugin encodes text on command line using %t in user's chosen encoding, but + if this differs from user's desktop default encoding, doesn't work. KProcess always + encodes command line stuff using user's default encoding. There is no point, therefore, + in attempting to encode it differently, just mangles the text even more. Short of modifying + KProcess, I see no way around this. But perhaps a warning to user advising to use %f instead? + _ Command plugin crashes when using command 'flite -t "%t"' and flite can't get write access to /dev/dsp*. +_ Hadifix plugin: + _ BUG:116151 Add options for using the new prepoc capability in Hadifix. + _ When stopText() is called, and Hadifix process is killed, the temporary wav file is not deleted, + despite QFile::remove call. Possibly still locked by exiting mbrola process? This only seems + to happen in the configuration dialog when canceling Tests. +_ KTTSMgr: + _ Give Talkers a Talker Name entered by user. Show this name in Jobs tab, Change Talker screen, + and in systray menu. + _ Change the Current Sentence panel to show the current sentence of the selected job? If selected + job is not the currently speaking job, won't show currently-speaking text anymore. However, would + be helpful when using advance/rewind buttons. Perhaps to be effective, if the currently-selected + job IS the currently speaking job, and that job finishes, would automatically select the next + speaking job? But that could become confusing for users with selection jumping around. + _ Add ability to quickly pick default Talker in systray menu. + _ KCMultiDialog has a bug that won't allow sizing the screen narrower than 640. + _ Overall Audio Speed should be a realtime setting, i.e., not necessary to click Apply. + _ Disable Audio Speed control if sox not installed (configure.in.in but runtime check would be better). + _ Add drag N drop for the Talkers so it is easier to change the ordering. + _ It is possible to configure multiple instances of Talkers with the exact same configuration. + _ In kcmkttsmgr, use setHelp() to set help anchor based on current tab. + _ Add playback controls to kttsmgr. Volume, pause. Maybe rewind? +_ KTTSD: + _ I am told that when kttsd crashes, it can take knotify down with it. See about + "disconnecting" knotify from kttsd (using post to Qt event queue). + _ Strange msg being produced by KIO::move, but it works fine. ?? + kio (KIOJob): stat file:///home/kde-devel/.kde/share/apps/kttsd/audio/kttsd-5-1.wav + kio (KIOJob): error 11 /home/kde-devel/.kde/share/apps/kttsd/audio/kttsd-5-1.wav + kio (KIOJob): This seems to be a suitable case for trying to rename before stat+[list+]copy+del + _ While kttsd currently configures a Talker automatically, would be nice if it could + autoconfigure talkers for each requested language, if one has not already been configured. + _ Add support for speech markers. + _ Implement error signal handling in KTTSD. + _ Add support for JSML and other Speech Markup languages. + _ Add method for apps to play a sound file, including in the middle of a job. + Could be done with SSML or with special Talker Code = "sound". +_ Other/General: + _ As pointed out by NOKUBI Takatsugu on Planet Debian, there exists an + open-source synthesis + recognition engine for Japanese. + http://aniki.daionet.gr.jp/~knok/blog/blosxom.cgi/2005/11/24 + http://hil.t.u-tokyo.ac.jp/~galatea/index.html + _ Save buttons should confirm if user chooses an existing file. KFileDialog doesn't seem to + offer this option, but shouldn't it? + _ .desktop files must have a Type keyword, but what to use for kcmkttsd_testmessage.desktop? + _ Add option to kill plugin process when no longer needed so that memory can be recovered. + Or maybe some sort of "nice" setting so that process will be swapped out when no longer needed. + _ Convert to KConfig XT. + _ Add a plugin for TuxTalk, http://tuxtalk.sourceforge.net/. Since it only supports + English (at the moment) and is alpha, don't think it is worth it now. + _ Add a plugin for dhvani, an Indian and Hindi TTS engine, http://dhvani.sourceforge.net/ + Currently written in c, but it looks like future versions will require Java VM. + Main problem at the moment is that dhvani sends audio directly to /dev/dsp with no option + to send to a wav file. + _ IBM ViaVoice plugin? IBM TTS is a proprietary runtime product with opensource API. + http://ibmtts-sdk.sourceforge.net/ + _ Add KTTS plugin to aKregator. |