The Slovene Interactive Text-to-Speech Evaluation Site (SITES) was built according to standards for interactive speech synthesiser comparison sites as set by COCOSDA (International Committee for the Co-ordination and Standardization of Speech Databases and Assessment Techniques for Speech Input/Output) and the LDC (Linguistic Data Consortium).
SITES aims to give the interested listeners a thorough and honest impression of the current text-to-speech (TTS) system and provides valuable feedback about strong and weak points of the system. The SITES web site enables to evaluate the S5 Slovene TTS system either interactively or off-line by sending the synthesized speech file to a given e-mail address. We implemented various standard text selection methods and set up rules for construction as Semantically Unapredictable Sentences for the Slovene language.
The evaluation web site has the capability to accept arbitrary input text, and returns a speech file. A CGI script first reads the user's form input. When the user submits the form, the script receives the form data as a set of name-value pairs, which is parsed. In the CGI script, the TTS system is called with the parameters specified by the user. The TTS system generates a temporal audio file which is sent back to the user.
[1]
Eric Moulines,et al.
Pitch-synchronous waveform processing techniques for text-to-speech synthesis using diphones
,
1989,
Speech Commun..
[2]
Nikos Fakotakis,et al.
Duration modelling for the greek language
,
1993,
EUROSPEECH.
[3]
France Mihelic,et al.
Speech timing in Slovenian TTS
,
1997,
EUROSPEECH.
[4]
Christian Benoît,et al.
An intelligibility test using semantically unpredictable sentences: towards the quantification of linguistic complexity
,
1990,
Speech Commun..
[5]
Tatjana Srebot-Rejec,et al.
Word accent and vowel duration in standard Slovene : an acoustic and linguistic investigation
,
1988
.
[6]
R.A.M.G. van Bezooijen,et al.
Assessment of synthesis systems
,
1997
.