README

                      Russian speech synthesizer

Ru_tts is a software Russian speech synthesis engine. It reads Russian
text in koi8-r from the standard input, performs TTS transformation and
writes the result to the standard output as raw linear signed 8-bit 10
kHz sound stream, which can be played by any appropriate sound player,
such as Sox, for instance. A simple example of such usage you can find
in the script ru_speak, provided with this package. By the way, this
script takes care of input text recoding. Thus, it accepts a text
in the current locale's character encoding and can be used for
immediate speaking. All command line arguments are passed to ru_tts.

Symbols '+' and '=' immediately after a vowel in input text
are treated as strong and weak stress sign respectively.

Ru_tts behaviour is controlled by command line options
that for convenience can be divided onto several groups.


Speech control options.

These options affect general speech parameters. The most of them
accept numeric value as an argument. All numeric values serve as
factors applied to a default level of the corresponding parameter.
Thus, they must be positive numbers and the default value is 1.0.
Greater value proportionally increases the parameter level,
lesser value proportionally decreases it.
Reasonable value range for various options is different,
but any out of range value will be automatically reduced
to the acceptable range.

-r value -- Set speech rate factor.
             Reasonable value range is [0.2..5.0].

-p value -- Set voice pitch factor.
            Reasonable value range is [0.5..3.0].

-e value -- Set speech emotionality factor.
            Reasonable value range is [0.0..1.4].

-g value -- Set interclause gaps duration factor.
            If this value is immediately prepended
            by one of the following punctuations ',', '.',
            ';', ':', '?', '!', the factor will be applied
            only to the gaps implied by that punctuation.
            If the value is prepended by symbol '-',
            it will be used for intonational gaps
            not caused by any punctuation.
            Otherwise all interclause gaps will be affected.
            Use this option several times to adjust various gaps.
            Lower boundary of the reasonable value range
            is always 0.0, but the upper one greatly depends
            on the value prepending symbol and actual speech rate.

-a -- Use an alternative (female) voice.


Numbers interpretation control options.

By default point and comma inside a number both are interpreted
as decimal separator. But this behaviour can be changed.

-d. -- Only point should be treated as decimal separator.

-d, -- Only comma should be treated as decimal separator.

-d- -- Disable decimal separators treating.
       In that case integer and fractional parts will be
       interpreted as two distinct integer numbers.


Pronunciation dictionary related options.

-s path -- Rulex pronunciation dictionary location.

-l path -- Log unknown words in specified file.
           It is intended mainly for dictionary debugging
           and is meaningful only in conjunction
           with pronunciation dictionary itself.


Other options.

-v -- Program name and version.

-h -- A brief usage description.