Commit Graph
35 Commits
Author SHA1 Message Date
jonsd 0fc12016fa [1.40.21]
Fix: -g command option not working on Windows.
Fix: Text position value in events was limited to 16 bits.

[1.40.15]
espeakedit:  Change formant dialog to include Klatt parameters.
espeakedit:  Change the format of formant data to include Klatt parameters.
espeakedit:  Fix the frequency of F7,F8 at 7800Hz,9000Hz.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@220 d46cf337-b52f-0410-862d-fd96e6ae7743
2009-05-29 11:48:39 +00:00
jonsd 0d3e23ef39 [1.40.07]
*_list files. Ad feature for ordinal numbers (for some languages).
*_list files. Allow hyphenated word groups to start with a number.
*_list files. Allow "words" of symbols to contain brackets (move check for bracket characters to after the *_list lookup).
--punct option. Fix some unnecessary pauses when speaking punctuation names.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@215 d46cf337-b52f-0410-862d-fd96e6ae7743
2009-01-18 11:37:10 +00:00
jonsd eabd70afbf [1.39.48]
Fix lettergroups: Sort items to longest first.

Lang en-us: Use rhotic [3] in place of [@] when [r] follows.

Voice files: add attribute "fast-test" to experiment with faster talking speeds.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@212 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-12-20 10:31:00 +00:00
jonsd 7758b77459 [1.39.42]
Languages which use non-Latin alphabets will automatically speak a word of Latin characters as English (or another specified language) without the need for entries in its *_rules file.

synthesize.cpp, SynthOnTimer():  Ensure the sound channel is open if there items still to be spoken, to prevent sound output lock-up.

Reduce warning messages from compilers.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@211 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-12-17 12:17:53 +00:00
jonsd 3096b7634d [1.39.39]
Windows SAPI5 installer: Remove the limit on the number of voices specified for inclusion in the Windows Registry.

voices.cpp:  Widen formant peaks F3 F4 F5.

Reduce slightly the difference in amplitude between stressed and unstressed vowels.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@208 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-12-12 13:33:52 +00:00
jonsd 897237bcf1 Phonetables: Use "length" field in Tone phonemes to adjust the lengths of the vowel.
git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@206 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-12-04 15:40:32 +00:00
jonsd f90a4e369e [1.39.36]
*_list files: add attribute $max3 to limit repetitions of a symbol to 3.
-v command option, if the voice-name is not found, lookup using it as a language name.
Windows: allow longer path names for espeak-data.
-m option: Ignore text within <script> and <style> tags.
Reduce default pitch to 80-117.
WavegenCloseSound() check if there remain items to be spoken and re-open the sound channel.
Reduce voicing before voiced plosives.
Corrections to mbrola translation for lang=hi.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@205 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-12-03 16:57:41 +00:00
jonsd 1150d1d8da [1.39.28]
Added experimental Klatt synthesizer.
Klatt phonemes use a new format of phoneme data.
espeakedit, new dialog fields and layout for Klatt phonemes.
Added Klatt versions of English phonemes.

Fixed some compilation warnings on Windows.
Fixed bug in Lettergroups > L10

Increased amplitude of harmonics below F1 peak, to increase bass.
Allow 5 (up from 4) pairs of values in "tone" command in voice files.
Default to slightly wider pk_shape1.

If words are separated by hyphens, don't add word-gap, and don't pass new-word flag onto phoneme processing.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@203 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-11-22 22:25:29 +00:00
jonsd a7501374b2 [1.39.26]
Added languages: ca (Catalan) lv (Latvian)
Numbers: allow for languages which have numbers for 100,000 and (or not) 10,000,000 (eg. hi, sw)
Phoneme [ts] new files in phsource/ustop
*_rules: fix crash for lettergroups numbers greater than 10, and better error reporting.
Languages: eo, jbo: adjust syllable lengths.
Language: en, fix "1st, 2nd, 3rd, 5th".  But a full ordinal number feature is still needed.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@202 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-11-16 09:18:53 +00:00
jonsd 084cd2c17d [1.39.22]
*_list files, allow words in all-capitals which imply $allcaps attribute.
Voices: Look for voices also in espeak-data/voices/test.
lang=eo: Adjust syllable lengths.  Use break between word (word_gap=1).
wave.cpp: Fixes to copyBuffer(), wave_write().


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@201 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-11-10 21:31:12 +00:00
jonsd 5ef05700e1 [1.39.07]
Fix  SSML <prosody rate> to allow the maximum rate of 390 wpm.
Fixes to the generation of the dict_phonemes report.
Lang=en-n adjust vowel sounds.
Compile phonemes: automatically assume vowel-break marker in second keyframe of a vowel file if none is set.
Set letter group 'A' to "vowel letters only" for Indic alphabets.



git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@195 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-09-27 18:14:15 +00:00
jonsd eab1faff96 [1.33] continued
git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@159 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-03-06 14:47:20 +00:00
jonsd 19db296d78 [1.31.14]
SSML: <voice>, if a language is specified, forget previously selected voice name.
Voice selection, convert language name to lower-case.
espeak-data/mbrola_ph files, allow use by both little and big-endian processors.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@149 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-02-22 13:59:25 +00:00
jonsd 39e8e1dcb6 [1.31.07]
intonation: changed comma intonation for intonation type 2.
intonation: adjusted exclamation (make the nuclear syllable emphasized).
intonation: split intonation table into body and nucleus parts.
lang-zh: only length tones 2,3 at end of sentence.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@137 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-02-08 10:38:45 +00:00
jonsd 703da28454 [1.31.05]
<say-as tts:char>, if a character name is not defined, say its English name.
phoneme tables, added beforenotvowel2 attribure for lang=tr
lang=tr: initial draft


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@135 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-02-05 01:18:56 +00:00
jonsd 38063c4eea [1.30.10]
Remove pause before <say-as> if it's at the beginning of a sentence.
Added --split=<minutes>  command-line option. Splits output WAV file (at sentence boundaries).
Fix: espeakEVENT_SENTENCE events were missing.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@121 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-01-13 14:45:58 +00:00
jonsd dd9fd5f27a [1.30.05]
Fix an error where vowels start too loud (too strong attack).

Added command-line option  --compile-debug, which is like --compile but it includes source line information from the *_rules file, which is displayed with the -X option. 

espeakedit: Added  Compile->Compile_dictionary(debug),

Fixed <emphasis> which was broken in the previous version(s).
<emphasis> now reduces the previous stressed syllable to secondary stress.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@119 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-01-11 00:36:01 +00:00
jonsd af575bc1db [1.30.04] SSML <audio src=""> now works without using a UriCallback() function, and works with the command-line espeak.
Pitch changes for mbrola are now applied over the first 80% of the phoneme, not over the whole length of the phoneme.  This seems to give better intonation for mbrola voices.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@118 d46cf337-b52f-0410-862d-fd96e6ae7743
2008-01-07 20:06:32 +00:00
jonsd f9725f9f64 [1.29.18] Add data for Serbian voice ("sr").
Moved Croatian/Bosnian/Serbian dictionary source data to hbs_rules, hbs_list.
Added data for Ancient Greek voice ("grc").
In voice files, the attribute: word n  adds a short pause between words. n = 1 to 4 for increasing pause length.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@103 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-11-10 22:10:43 +00:00
jonsd 531de957cd [1.29.15] Fix compiler warning, conversion from string constant to 'char *'.
espeakedit: Change compile_report file to include the phoneme name as well as phoneme table which uses each data file.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@95 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-11-01 19:09:27 +00:00
jonsd 75f0402750 Remove duplicate definition of wave_flush() in wave.cpp.
git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@81 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-09-01 10:31:06 +00:00
jonsd 3d97adb438 [1.29] lang=en, minor changes to phonemes [I] [V] [aI] [eI]
git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@80 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-08-27 12:57:24 +00:00
jonsd cbff131b89 [1.28.07] Changed the format of envelope files (phsource/envelope). These are not text files which contains coordinates which define the envelope.
The compile_phoneme_data operation now produces an image file phsource/envelopes.png which shows the shape of the envelopes.

git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@78 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-08-21 14:44:24 +00:00
jonsd 5019e65842 [1.28.05] Changed names of voice variant files (espeak-data/voices/!v).
Moves "wisper" and "croak" into voices/!v.
Voice variants can now be specified as  en+wisper, en+f2 (for female 2), etc.
espeak --voices=variant  lists the voice variants.
Moves some lines from speech.h into other header files.
Possible fix for data version check on big-endian.
Fixed crash error iin LoadConfig().
Fixed crash in espeakedit: press Tab in enpty Prosody window.

git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@74 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-08-17 12:58:34 +00:00
jonsd 810fb2404d [1.28.03] Added macro in speech.h for max. length of the espeak-data path.
Added language: "bs" Bosnian.
Fix to speak_lib.h for Windows dll version.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@69 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-07-31 09:17:13 +00:00
jonsd 0def252307 [1.27.09] Fix for --punct option when used with <say-as>.
lang=hu, shorten "ll". Test: remove differences in vowel lengths depending on the following phoneme.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@64 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-07-12 21:39:42 +00:00
jonsd 1a28202d00 [1.27.04] tr_languages: replace_char substitution can do 2->1 and 1->2 characters.
espeakedit: set and save path for master phonemes file.
espeakedit: added delete_phoneme and import_phoneme actions in phoneme files.
espeakedit: improve data in log_espeakedit, include phoneme names.
espeakedit: only write to log_espeakedit for speak from text window, not when playing vowel files.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@59 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-07-04 22:59:40 +00:00
jonsd 05aeea7858 [1.27] Only keep one primary stress when adding prefixes.
Add stress_rule=4, altepenultimate stress.
Change license to GPL version 3.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@56 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-06-30 12:31:21 +00:00
jonsd 3ab4a32bab [1.26.03] Priority stress [''] reduces other primary stress to secondary not unstressed (except lang=pt).
Add short pause between [*] and a fricative.

git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@44 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-06-12 11:49:46 +00:00
jonsd 85cb0ec378 [1.25.14] Added feature to define collections of letter sequences in *_rules files for use in rules.
lang=pt  Reduce phoneme [&~] to [&] in unstressed syllables.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@30 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-05-31 11:13:24 +00:00
jonsd ec9796cb99 [1.25] Enable exclamation intonation for lang=pt
Lang=fr, [o],[O] more open.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@15 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-05-19 01:24:45 +00:00
jonsd f65bf2072b [1.24.12] Produce dictsource/dict_phonemes, a list of which phonemes are used by each language's *_rules and *_list files.
Fix crash with -X option when spelling words (acronyms).
Rules files: added special character X meaning "no vowel until the word boundary". Used for lang=no.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@14 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-05-18 10:52:23 +00:00
jonsd 6d5a152841 Fix voice attribute suffix ignored in voice name.
lang=cs   fix number not spoken after preposition with $alt1 attribute
lang=cs   Don't allow any thousands separator.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@12 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-05-17 00:58:01 +00:00
jonsd 3899776fad Write version_phdata number to phondata file so that the byte order is correct for both little and big-endian processors.
Adjust phoneme [R].
Lang=ru  Revert to phonemes [S] and [Z] instead of [s.] and [z.].
Lang=ru  Remove rule to give 1st syllable a secondary stress if the second syllable has a primary stress.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@9 d46cf337-b52f-0410-862d-fd96e6ae7743
2007-05-12 17:29:43 +00:00
jonsd 8d59235f7a git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@1 d46cf337-b52f-0410-862d-fd96e6ae7743 2007-05-05 11:00:27 +00:00