Akademik Veri Yönetim Sistemi
International Journal of Humanities Social Science and Management, cilt.4, sa.4, ss.109-121, 2024 (Hakemli Dergi)
The sudden increase of the workload
and transaction speed in daily life forced humans to
develop new technologies to adapt these situations.
One of these technologies is the complex software
that transform text into speech. This study aims at
developing a phoneme/phone-based text-to-speech
software, which does not need large databases, and
which can run on different platforms. To achieve
this, the phonological structure of Turkish was
investigated and then a phoneme/phone-based textto-speech software is developed, differently from
most of the previous studies. At the end of the study,
a software that runs with a small database is
developed. Since the rules, symbol and recordings
used in the algorithm are kept separately from the
code of the software, it allows to change these data
without altering the code. Thus, it would not be
necessary to write the code anew for future
improvements. The powerful morphology of
Turkish, and the complex morphology-phonology
interaction prevented processing of prosodic
features such as stress and intonation. To perform
such a processing, a large and annotated database is
needed to conduct morphological and syntactic
parsing. Considering these limitations of our study
we believe that the software we developed runs
smoothly, however it requires improvement
regarding prosody.