Example for number reading after fix:
23 -> esR,imv,eSal'oS
23.4 -> esR,imv,eSal'oS nekud'a aRb'a?
23.45 -> esR,imv,eSal'oS nekud'a aRba?,imv,eXam'eS
23.456 -> esR,imv,eSal'oS nekud'a aRb'a? Xam'eS S'eS?

Set LOPT_APOSTROPHE bit 1 for Hebrew to ensure final apostrophes are treated as part of the word. This is relevant for words like בָּגָאז׳ or חָגָ׳ג׳, where the apostrophe represents a phonemic element and should not be ignored during tokenization or phoneme conversion.

Replacement rules will change  ׳ (Hebrew GERESH = U+05F3) to ' (standard apostrophe)

The threshold value for the activation of Sonic was changed in 065a0e8f03 without an obvious reason.

Related issues: #131, #1461.

This commit adds support for the Pashto language (ps) to espeak-ng. The implementation is based on Urdu language files and includes ps_rules, ps_list, ps_emoji, ps_extra files and updated Makefile.am to include Pashto in the dictionary targets and build rules.

fix: add Pashto data file.

refactor: use enhance rules for stress.

fix: add missing configs.

Add Pashto phoneme support and improve voice files

fix: add Pashto phonemes test.

fix: restore original phonemes.

fix: remove renduandant ps_dict from Make.am file.

fix: use correct phonemes with ipa and stress rules.

feat: translate all en_emoji to Pashto.

fix: add Pashto dict entry in Makefile.am

feat: enhance ps_rules with example pairs and words.

but the user requests it

Ref: #1807

The implications of this are large for using eSpeak-ng as a phonemizer. It means that it's now possible to get accurate mappings between the phonemized word and the original word (up till now it was pretty much impossible). This is huge for many applications that use it as a side-tool.

This reverts commit 56a4c82d99.

`#if PLATFORM_WINDOWS` doesn't work for some reason.

Fix suggested by @beqabeqa473

So that SelectVoice can select an mbrola language.

Fixes #1011

Otherwise we won't ever be able to select an mbrola voice in ssml
processing.

We should initialize current_voice_id so that ssml processing knows what
voice we currently have, and not try to change it unless really needed.

When e.g. using mbrola voices, the sample rate may change. When the
application plays audio itself, it needs to know it, and
espeak_ng_GetSampleRate does not report that.

this fix was lost in commit 84e967c0a, I can’t even imagine how it happened.
thank God it’s not me who was merging it.