brailcom/speechd
View on GitHub`GenericMaxChunkLength` causes incomplete multibyte characters
Open
#806 opened on Jan 18, 2023
buggood first issuehelp wanted
Repository metrics
- Stars
- (325 stars)
- PR merge metrics
- (PR metrics pending)
Description
Steps to reproduce
Create /etc/speech-dispatcher/modules/test.conf
# GenericMaxChunkLength 300
GenericExecuteSynth "echo $DATA >> /tmp/tts.log"
GenericLanguage "en" "en" "utf-8"
GenericLanguage "zh" "zh" "utf-8"
AddVoice "en" "male1" "voice1"
Add configuration to /etc/speech-dispatcher/speechd.conf
AddModule "test-generic" "sd_generic" "test.conf"
Restart speech-dispatcher and use test-generic
$ sudo killall speech-dispatcher
$ spd-say -o test-generic "ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇"
Obtained behavior
$ cat /tmp/tts.log
ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇〇〇〇〇〇〇〇ab〇〇〇�
��〇〇〇〇〇〇
Passing incomplete multibyte characters to external commands is likely to not work properly, and subsequent chunks will not even be recognized as UTF-8 strings.
Expected behavior
Do not break down multibyte characters.
The string should be broken down into chunks of the specified length in characters (not bytes).
Behavior information
Distribution
Debian
Version of Speech-dispatcher
0.11.4