Conflicts:
	README.md

The i phoneme is used for the reduced i sound in e.g. happy
(/h'api/). Thus, the following words contrast:

    word   eSpeak   IPA
    happy  /h'api/  /hˈapi/
    been   /b'i:n/  /bˈiːn/
    bin    /b'In/   /bˈɪn/

Before this change, /h'api/ would be transcribed in IPA as /hˈapɪ/
which is wrong. After this change, it is transcribed correctly.

When running any of the following:

    espeak -x "[[]]"
    espeak --ipa "[[]]"

eSpeak would SEGFAULT. This is because phon_out_buf is NULL as no
phonemes have been written and as this is the first call to
GetTranslatedPhonemeString, the phoneme buffer has not been
allocated yet. However, GetTranslatedPhonemeString attempts to add
a NULL character to the end of the phoneme buffer, and thus crashes.

Android does not support UN M.49 territory codes. Instead:
  *  4.2 and earlier return an empty string;
  *  4.3 throws a MissingResourceException.

This is used to filter the list of languages returned. However, the
filtering is not needed on the engine side as the client can simply
enumerate through all the available voices instead.

Running |espeak --compile=om| on some platforms results in espeak
segfaulting intermittently (not reproducible under gdb). When it
does, it generates "Unknown keyword: " messages with garbage after
it.

Conflicts:
	dictsource/dict_phonemes
	phsource/compile_report

Language updates: da, en, nl, pt, vi


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@363 d46cf337-b52f-0410-862d-fd96e6ae7743

Conflicts:
	src/speak_lib.h

git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@362 d46cf337-b52f-0410-862d-fd96e6ae7743

Conflicts:
	dictsource/dict_phonemes

Change API function: espeak_TextToPhonemes().
Fix bug: number following a non-ascii character could cause crash.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@361 d46cf337-b52f-0410-862d-fd96e6ae7743