Wait for the process to signal new data available instead of polling its
output. Fixes synchronisation issues with sapi_voice.vbs that were causing
FS#12595.
Change-Id: I86e844e837ed20f32005fdccf7ec9e8c1efbe983
Instead of simply assuming the wav file that is supposed to be created by the
TTS engine check if the file actually exists and return an error if not.
Change-Id: I9e4a85a061b44b48931614602683b1dfe7dfce67
Support retrieving the vendor name of the TTS. This will be used by TTS string
corrections. Currently no other TTS but SAPI supports this, and only correction
strings for SAPI voices depend on the vendor information.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@30609 a1c6a512-1295-4272-9138-f99709370657
The settings currently can't differentiate between default values for different
subitems (i.e. different TTS systems). As a result the default value returned
for SAPI speed was way out of bounds, leading to the upper limit getting used.
Work around this limitation by checking the value for bounds and use a sensible
value in that case.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@29763 a1c6a512-1295-4272-9138-f99709370657
- Use ISO codes as keys for voice languages instead of enumeration.
- Instead of trying to select a suitable voicefile language from the UI
language use the ISO codes stored as UI language and match against the
mapping.
- Always store the selected UI language. Fixes the voicefile creation
language lookup to fail if the language used is the system language.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@29207 a1c6a512-1295-4272-9138-f99709370657
