A program and a library that can be used as a simple
basic interface to some SoX functionality or for producing
the approximately Ukrainian speech with your own recorded
voice (actually it produces the needed sound representations).
The program starts with Caution to be responsible for usage
and to use it personally. Then the program guides you
through the creating and using your Ukrainian "voice".
Please, use it carefully. After the execution you can
execute mmsyn7h program to produce some records with your
newly created "voice". During the execution of the last one
you can choose whether to remove (clean)
all the created sound files in the current directory for
security reasons. If the programs terminate by interruption (or
generally) it is a good practice to remove the current
directory sound files manually.
***** CAUTION! *****
"The possession of great power necessarily implies
great responsibility."
(William Lamb)
The program mmsyn7ukr produces the "voice" represented as an ordered
set of sounds each of which corresponds (represents) one of the
Ukrainian sounds so that using them together by mmsyn7h program
(https://hackage.haskell.org/package/mmsyn7h) can be a background
for sound synthesis. If you pronounce sounds as the appropriate
Ukrainian ones, close to proper Ukrainian speech with your own
voice. This program approximates your voice with a sequence
of recorded separate sounds with your control over the
duration of the sounds. They are then precessed by the SoX
binaries already installed in the system to produce the
needed sounds and then you can pronounce some Ukrainian text
with your recorded "voice" using mmsyn7h program. In-between
you can do some additional processing as you need. Moreover,
you can substitute whatever sounds you like (consider being
sensible) instead of your own voice.
Be aware that if somebody can get access to the sounds of
your voice or to the recorded speech (except you) then this
possibility itself creates security issues and concerns. So,
please, do NOT give access to such records to anybody else
except you.
In such a case, the program is for personal usage of every
user ONLY!
Being given as an advice in such a case, run the program
in the empty directory with the current user permissions to write,
read and search and provide some proofs and evidence that nobody else
can even read the files in the directory. May be, it is better
to execute the program being in the directory located in the RAM,
then consequently wait until the program ends and then reboot
the computer.
If the program ends earlier, you must then remove
(better wipe) the directory contents. No other users should
have access to the computer after you have begun to run the
program and have not deleted (or better wiped) the contents
of the directory. Please, be aware, that there are possibilities
to read sensitive information from the drives after you have
deleted the files in a usual way. You can use wiping for better
security. Besides, if somebody can get access to the memory of
the computer or to the directory contents where you run the
program or (may be) to the temporary files created by SoX or
to the drive where you run the program (not in the RAM, or may
be in it) then your voice can be stolen and / or used
inappropriately. Use all possible precautions and measures to
avoid the situation.
Be aware also that the given by the program technology (or
documentation for it in any form) of the voice processing can
be improved so there is NO guarantees that the given technology
or its successors cannot be used in violating your voice identity
to produce from some available voice records the voice for the
inappropriate usage. Therefore, better is to prove your identity not
only with the solely voice itself but with some additional
independent sources and measures.
The author of the program accordingly to the LICENSE (MIT) does not
response for any possible issues, but by this notification tries to
intent you to be aware of some possible issues.
***** Command Line Arguments *****
If you specify as a first command line argument one of the numbers below
the program behaves as follows:
-1 -> the program does not reduce noise, it only resamples
the audio to the needed 22050 Hz and adjusts the amplitude;
0 -> after the noise reduction the program only resample
the audio to the needed 22050 Hz and adjusts the amlitude;
1 -> after the noise reduction the program additionally
to the 0-processing truncates the silence from the beginning
and end of the audio to the level given by the second
command line parameter;
2 -> after the noise reduction the program additionally to
the 1-processing applies a double band-reject filter to
the audio (SoX "sinc" effect);
3 -> after the noise reduction the program additionally to
the 2-processing applies fade-in and fade-out effects to
the audio;
_ -> is the same as 3.
If you specify as a second command line argument one of the numbers below
the program behaves as follows:
0 -> the maximum amplitude, starting from which the file will
not be trimmed for the first command line argument greater
of 1, is 0.01;
1 -> the maximum amplitude, starting from which the file
will not be trimmed for the first command line argument
greater of 1, is 0.02;
2 -> the maximum amplitude, starting from which the file
will not be trimmed for the first command line argument
greater of 1, is 0.04;
3 -> the maximum amplitude, starting from which the file
will not be trimmed for the first command line argument
greater of 1, is 0.08;
_ -> the maximum amplitude, starting from which the file
will not be trimmed for the first command line argument
greater of 1, is 0.04;
If you specify the third command line argument, it must be a list
of Strings that can be obtained by executing the mmsyn7s program
(https://hackage.haskell.org/package/mmsyn7s). The program will create
just that non-silent representations for the Ukrainian sounds,
which are given. The list must be sorted.
***** More Information *****
You can create a sound in either a 'sharp' mode or in a usual mode.
The first one means that the program does not check whether the
specified duration for recording the initial sound data
is greater than 3 seconds. In such a case the duration can be
much smaller. This mode needs more mastership in interacting with a
program. For speech synthesis (as an advice) use this mode for
the very short sounds (like the sound representation for "ь")
or for the sounds, you can articulate for a long time continually
(for example, vowels and some consonants). The 'sharp' mode delegates
the responsibility for the sound to much broader extent to the user,
so for a beginning it is not recommended (though you can give it a try).
At the beginning the program also creates a noise profile (once per execution).
It is now used to reduce the noise level for the recorded sound representations.
It uses the default SoX noise reducing settings with a hope that for you they can
be sufficient. If you see that the program removes some sensitive and
important sound data from the recorded by it sound representations,
then you can specify as a first command line argument "-1". The program
in such a case will create a noise profile as usual, but will not reduce
the noise at all.
***** Ukrainian Localization *****
Please, before using the program check that uk_UA.UTF8 localization is
present in your system as one of the system locales. Otherwise,
the program will possibly (in some cases surely) cycle. In such a case,
you can terminate it in a usual way by sending interruption signals.
***** SoXBasics and SoXBasics1 *****
The library have two similar modules with partially overlapping function names
(so it is recommended to import them qualified). SoXBasics1 functions after
creation of the resulting file try to replace by it the initial one and to clean
the space from the temporary files. This can be helpful in writing sequentially
applied functions but needs somewhat more resources.