TGSpeechBox v3.0 beta 7 out for you all
-
TGSpeechBox v3.0 beta 7 out for you all!
The headline is the full pronunciation dictionary system, now on every platform: iOS, Android, and the Windows phoneme editor. Four dictionary types — pronunciation, stress, compound, and character — with add, edit, delete, search, and import/export. The Windows editor adds "Convert to IPA" buttons that call eSpeak to pre-fill entries. Fully accessible with TalkBack and VoiceOver.
SAPI users finally get real engine control: a full settings panel covering sample rate, all 14 voicing tone sliders, voice quality parameters, pitch mode, inflection scale, and pause mode. Settings persist in AppData. This brings SAPI in line with what mobile users have had for a while.
Under the hood: ~900 lines of duplicated acoustic math between emitFrames and emitFramesEx merged into a single template, eliminating a whole class of ghost bugs where a fix in one path never reached the other. The hard-knee limiter is replaced with a soft-knee design — Android users at high gain will hear the difference immediately.
Bug fixes: phoneme edits on mobile now actually reach VoiceOver and TalkBack (they weren't before), locale-sensitive number parsing fixed for Hungarian and Polish, and dictionary text fields no longer autocorrect your IPA.
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSpeechBox-v300b7.nvda-addon
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSBPhonemeEditor-v300b7.zip
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSpeechSapiSetup-v300b7.exe
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSpeechBox-v300b7.apk
https://testflight.apple.com/join/jvvGY6Fz
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/tgspeechbox-linux-x86_64-v-300b7.tar.gz
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/tgspeechbox-linux-aarch64-v-300b7.tar.gz -
TGSpeechBox v3.0 beta 7 out for you all!
The headline is the full pronunciation dictionary system, now on every platform: iOS, Android, and the Windows phoneme editor. Four dictionary types — pronunciation, stress, compound, and character — with add, edit, delete, search, and import/export. The Windows editor adds "Convert to IPA" buttons that call eSpeak to pre-fill entries. Fully accessible with TalkBack and VoiceOver.
SAPI users finally get real engine control: a full settings panel covering sample rate, all 14 voicing tone sliders, voice quality parameters, pitch mode, inflection scale, and pause mode. Settings persist in AppData. This brings SAPI in line with what mobile users have had for a while.
Under the hood: ~900 lines of duplicated acoustic math between emitFrames and emitFramesEx merged into a single template, eliminating a whole class of ghost bugs where a fix in one path never reached the other. The hard-knee limiter is replaced with a soft-knee design — Android users at high gain will hear the difference immediately.
Bug fixes: phoneme edits on mobile now actually reach VoiceOver and TalkBack (they weren't before), locale-sensitive number parsing fixed for Hungarian and Polish, and dictionary text fields no longer autocorrect your IPA.
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSpeechBox-v300b7.nvda-addon
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSBPhonemeEditor-v300b7.zip
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSpeechSapiSetup-v300b7.exe
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSpeechBox-v300b7.apk
https://testflight.apple.com/join/jvvGY6Fz
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/tgspeechbox-linux-x86_64-v-300b7.tar.gz
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/tgspeechbox-linux-aarch64-v-300b7.tar.gz@Tamasg I started beta testing this on testflite yesterday on my mac, and I also left you some feedback on it.
I'm absolutely loving this TTS.
Thank you so much for creating this. -
TGSpeechBox v3.0 beta 7 out for you all!
The headline is the full pronunciation dictionary system, now on every platform: iOS, Android, and the Windows phoneme editor. Four dictionary types — pronunciation, stress, compound, and character — with add, edit, delete, search, and import/export. The Windows editor adds "Convert to IPA" buttons that call eSpeak to pre-fill entries. Fully accessible with TalkBack and VoiceOver.
SAPI users finally get real engine control: a full settings panel covering sample rate, all 14 voicing tone sliders, voice quality parameters, pitch mode, inflection scale, and pause mode. Settings persist in AppData. This brings SAPI in line with what mobile users have had for a while.
Under the hood: ~900 lines of duplicated acoustic math between emitFrames and emitFramesEx merged into a single template, eliminating a whole class of ghost bugs where a fix in one path never reached the other. The hard-knee limiter is replaced with a soft-knee design — Android users at high gain will hear the difference immediately.
Bug fixes: phoneme edits on mobile now actually reach VoiceOver and TalkBack (they weren't before), locale-sensitive number parsing fixed for Hungarian and Polish, and dictionary text fields no longer autocorrect your IPA.
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSpeechBox-v300b7.nvda-addon
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSBPhonemeEditor-v300b7.zip
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSpeechSapiSetup-v300b7.exe
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/TGSpeechBox-v300b7.apk
https://testflight.apple.com/join/jvvGY6Fz
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/tgspeechbox-linux-x86_64-v-300b7.tar.gz
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300b7/tgspeechbox-linux-aarch64-v-300b7.tar.gz@Tamasg Whatever happened to this since the last build I was running, I like it. American still doesn't quite sound right to me; the way it says a lot of words with an r in them (like forty, which I heard a lot when I was playing with settings) feels off, and it just seems like it's having a bit of an identity crisis. I usually use British ESpeak and that carried over to here, so it took me a while to notice.
-
@Tamasg Whatever happened to this since the last build I was running, I like it. American still doesn't quite sound right to me; the way it says a lot of words with an r in them (like forty, which I heard a lot when I was playing with settings) feels off, and it just seems like it's having a bit of an identity crisis. I usually use British ESpeak and that carried over to here, so it took me a while to notice.
@simon fixed! Thanks for bringing more attention to "four" and "fourty." There were 3 bugs in the YAML that stopped the correct rule from applying! Fixed here: Https://eurpod.com/en-us.yaml
If the page opens in-browser, you can right click the text and do "save page as" (Firefox) or save it from Chrome's menu.
Once you have it, paste ithere: %appdata%\nvda\addons\tgspeechbox\synthdrivers\tgspeechbox\packs\lang
Done. no new add-on needed, just a simple YAML data change. -
R relay@relay.infosec.exchange shared this topic