Use the following tools

    Mbrola:  .pho files -> .wav files

    FreeTTS: text -> .pho files, then calls Mbrola

    Lame:    .wav -> .mp3

    ttsprep: Prepares text files for FreeTTS.

MBrola:

    Install the binary system (will need path below).
    Install voices us1, us2, us3.  The language binary files must
     reside at locations like $MBROLA_HOME/us1/us1.

FreeTTS:

    Sym link $FREETTS_HOME/mbrola/mbrola.jar to $FREETTS_HOME/lib

    java -cp /usr/local/voice/FreeTTS/classes -Dmbrola.base=/usr/local/voice/mbrola com.sun.speech.freetts.FreeTTS -file /tmp/soab.txt -voice mbrola2
    java -cp /usr/local/voice/FreeTTS/classes -Dmbrola.base=/usr/local/voice/mbrola com.sun.speech.freetts.FreeTTS -file ortho01.mid -voice mbrola2 -dumpAudio ortho01.wav
    (Version 1.2 of FreeTTS not working, at least not the instructions for
    java -jar ../freetts.jar... and java -jar .../mbrola.jar).

Lame:

    lame --tt 'Chapter 1' --ta 'G.K. Chesterton' --tl Orthodoxy --tg Speech --ty 1908 --tn 1 ortho01.wav ortho01.mp3



txt2wav = ttsprep + FreeTTS:
    
    ttsprep file.txt |
    java -cp /usr/local/voice/FreeTTS/classes -Dmbrola.base=/usr/local/voice/mbrola com.sun.speech.freetts.FreeTTS -voice mbrola2 -dumpAudio file.wav


SPLITTING input text files
    java -DFILLOUT=3 BLSplit $HOME/mp3/aloj/aloj.txt 3000
    That would write files to $PWD of size >= 3000, broken on blank lines,
     with names like aloj001.txt, etc.

txt2waving a set of files:
    for file in aloj???.txt; do  txt2wav $file || break; echo -e -n .; done; echo