bugfix: allow both SSSML <audio> and soundicon punctuation
All sound file information goes to soundicon_tab. The contents of the files are not
loaded to memory before they are needed for synthesis.
This removes N_SOUNDICON_SLOTS that divided soundicon_tab into two
parts for unknown reasons.
TODO: LoadConfig() is currently undocumented. The logic should be
moved to LoadVoice() and documented.
Soundicons are used for external audio with SSML <audio> tag and for
replacing punctuation names with sound files in LoadConfig().
Currently there's a bug wih soundicon slots: if both LoadConfig and
<audio> are used, the punctuation reserves all slots and no sound from
<audio> is played.
The SelectPhonemeTable(voice->phoneme_tab_ix) has no effect for tests.
Passing other values here breaks tests. That means that the relevant
calls SelectPhonemeTable() are already made in other functions.
voice->phoneme_tab_ix is set in LoadVoice().
voice fast is related to speed.fast_settings. Others were errors in
LoadVoice() parsing or commented out code.
Note that some voices are from the original espeak and have comments
that might be outdated.
speed.fast_settings[] was introduced in 7b2eecc724 to be used with
Sonic. Looks like after 7b2eecc724 the code related to everything
except fast_settings[0] was removed.
Might be that speed.fast_settings[0] is also irrelevant and could be
removed.
See commit message for 23a4d88f.
This commit fixes cmn and yue.
CalcPitches_Tone() now accepts cmn for translator_name.
SelectTranslator() now has a case for yue instead of zhy.
Option "language <name>"already causes SelectTranslator(<name>) to be
called. Having two options to do almost the same thing is unnecessary
and confusing.
In the long term, all options from SelectTranslator() should have a
switch case in LoadVoice() so they are user configurable (see #218). If
needed, a new option (maybe called "LoadOptions") could be added to load
an existing voice or language file.
Changes language configuration files for: hak, cmn, yue, ltg, ms, mb-ma1.
No changes to users.
voices: Change default number pronunciation rule to enabled.
docs: add details about number flags to the documentation.
It's clearly intended to be enabled by default:
- it's defined as default behaviour translate.h (NUM_DEFAULT)
- tr_languages.c sets many default values related to number processing
that have no meaning unless langopts.numbers == 1.
It is also a more sensible default since most languages will want to
have number processing on. This makes adding new languages easier
because adding an entry to tr_languages.c is unnecessary.
A negative side effect is that languages with partial number defines
might experience bugs when reading undefined numbers. This is a bug and
should be fixed.
This will have the side effect of enabling number processing for
languages that currently have it disabled. However, there shouldn't be
any.
Here's a way to check affected languages:
for voice in $(ESPEAK_DATA_PATH=`pwd` LD_LIBRARY_PATH=src:${LD_LIBRARY_PATH}
src/espeak-ng --voices | grep -v Languages | awk '{print $2}'); do
OUTPUT=$(ESPEAK_DATA_PATH=`pwd` LD_LIBRARY_PATH=src:${LD_LIBRARY_PATH}
src/espeak-ng -qx -v $voice "1 - 2 - 3 - 12 - 123") && echo "$voice:
$OUTPUT" ; done
These voices clearly benefit from enabling numbers (they already have
number rules in *_list):
ba, cmn (zh), hak, haw, ja, kok, nb, nci
Some languages are missing some definitions (like _12) in _list files.
It causes the program to skip some numbers.
Numbering needs to be turned off explicitly for:
jbo, mi, my, piqd, py, qu, quc, th, uz
Languages with no number rules at all:
chr, cv, he, nog, tk, ug
Espeak-ng uses bitrate 16 internally. If the input inside <audio> has a
different bitrate the audio will not play correctly unless converted to
16 bits.
Closes #885.
It is supposed to be used with SSML for reading < (<) and & (&)
when invoking espeak-ng --punct -m.
However, it looks like the code is dead since removing it doesn't fail
the ssml reference test.
using sox for ssml <audio> tag is hardcoded in readclause.c and for
--compile-phonemes in compiledict.c
Currently, --compile-phoneme doesn't need to run sox since the
samplerate for the specific file is correct. There might be a lot of
dead code in compiledata.c
code cleanup: remove unnecessary current_voice_id in readclause.c
Instead of passing a temporary variable current_voice_id to SSML
processing logic and then copying the value to voice_change, pass
voice_change directly. GetVoiceAttributes() both sets voice_change and
retuns CLAUSE_TYPE_VOICE_CHANGE that is used by TranslateClause() for
error checking.
tests/ssml/language-switch.ssml covers the intended behavior of language
switching and will fail when either voice_change or CLAUSE_TYPE_VOICE_CHANGE
is not set correctly.
katrihiovain