A couple quick points.

There are two places where espeak-ng handles stop requests. One is in close_stream, the other is at the bottom of the main loop in say_thread. As far as I can tell, the one in close_stream is rarely or never exercised. So I patched fifo.c to call audio_object_flush in say_thread.

Second, in speech.c, audio_object_flush gets called in espeak_ng_Cancel. This call happens in a thread other than say_thread, which is the one doing the audio handling. pcaudiolib isn't guaranteed to be threadsafe, so I don't think we can call audio_object_flush there at all. What is that going to do to responsiveness?

There are two branches on my fork https://github.com/CMB/espeak-ng. One is stoprequest, which adds audio_object_flush to say_thread. The other branch is named threadsafe, and it has a patch to remove the flush call from speech.c as well as the stop request patch.

[pcaudiolib:master] By Reece H. Dunn <msclrhd@...>:
8a013933f66b: alsa: fix audio echos when flushing ALSA devices

Modified: CHANGELOG.md
Modified: src/alsa.c

[espeak-ng:master] New Comment on Issue #172 sluggish speech interruption
By rhdunn:

I have pushed a patch to pcaudiolib to do the close/open logic in there (in the ALSA device), instead of using a workaround in espeak-ng.

Hi Samuel,

Thanks for investigating this further. I have pushed a different implementation of my original fix and your fix-cancel.txt patch to call audio_object_flush instead of close/open. The audio_object_flush implementation will be fixed to resolve this issue.

is it just ALSA that is the issue, or is pulseaudio affected as well?

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
dee3799f82bc: issue #172: flush audio on cancel requests. This is a partial fix for an audio regression. The other part of the fix requires a change in pcaudiolib to reopen the audio device on flush requests for affected audio drivers/devices.

Modified: src/libespeak-ng/fifo.c
Modified: src/libespeak-ng/speech.c
Modified: src/libespeak-ng/speech.h

Hello,

I have dug a bit more into the issue our users are getting, and it happens that it seems like a bug in ALSA, in that snd_pcm_drop doesn't actually drop the PCM data from buffers, see for instance http://stackoverflow.com/questions/15826155/alsa-snd-pcm-drop-is-not-clearing-complete-buffer . Some users have indeed reported real overlapping of speech: the last bits of PCM is simply restarted along the new bits of PCM... This is possibly dependent on the actual ALSA driver.

What espeak was doing which avoided the issue was to actually shut down the PCM object, and that indeed manages to completely flush the ALSA buffers and thus avoids the issue. I have attached here fix-cancel.txt the corresponding fix. It brings back part of f9ab812 : reintroduce close_audio, but also a new open_audio, so that the cancel code can close/open the audio stream to make sure it gets flushed.

I have left a FIXME in the patch: there is still one case which is not fixed, it's when one wants to cancel the very end of an ongoing speech. In that case espeak-ng has finished producing the samples and has pushed the last bits to ALSA and is just blocked inside the audio_object_drain() call. If the user calls cancel here, the ongoing audio will not be interrupted. Depending on the ALSA setup, that might be more or less long. Instead of just draining, espeak-ng should instead drain with a timeout, so as to get back control periodically in order to check my_stop_is_required from times to times (e.g. 60ms). That's however not supported by pcaudiolib yet, so it's not a simple fix, I won't be able to do that for Debian Stretch.

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
c36789e8f7d6: docs/phoneme_model.md: create a new phoneme model document.

Added: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
bf389d8d6252: docs/phoneme_model.md: move the manner of articulation details to this document.

Modified: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
aebcafa19d88: docs/phonemes.md: make the manner of articulation table consistent with the other feature tables.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
7eb3abee6e7b: docs/phonemes.md: move the IPA phoneme sections up a level.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
4f894a65fb4e: docs/phonemes.md: move the manner of articulation into the consonants section.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
4d750ed4d851: docs/phonemes.md: restructure the articulation diacritics.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
58d97b0c8882: docs/phonemes.md: move Air Flow to the diacritics.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
43f72c5e527d: docs/phonemes.md: restructure the phonation features.

Modified: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
7a707072ded9: docs/phonemes.md: restructure the place of articulation features.

Modified: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
9c18c31d61ae: docs/phonemes.md: restructure the rounding and labialization features.

Modified: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
6ee32cbac1be: docs/phonemes.md: restructure the vowel height and backness features.

Modified: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
e4daf9fbb7dd: docs/phonemes.md: restructure the syllabicity features.

Modified: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
c6442f2b7d33: docs/phonemes.md: restructure the consonant release features.

Modified: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
896108467d15: docs/phoneme_model.md: add place of articulation to the table of contents.

Modified: docs/phoneme_model.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
7199046f9580: docs/phonemes.md: restructure the co-articulation features.

Modified: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
2d2cabed66cc: docs/phonemes.md: restructure the remaining diacritic features.

Modified: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
f7630893bf38: docs/phonemes.md: restructure the suprasegmental features.

Modified: docs/phoneme_model.md
Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
ae311b0c2888: docs/phonemes.md: update the opening paragraphs to reflect the scope of the document.

Modified: docs/phonemes.md

Oh, thats my fault. When I compiled it successfully, I didn't expect, that actually wrong file was used. I made comment it the mailing list about this issue.

Min,

I merged `phsource/ph_myanmar` faulty, because espeak-ng uses `ph_burmese` name. And, when Reece did it,

he got compilation errors, because your project doesn't have  referenced phoneme files (`my` and `FinalMy` folders) in github repository.

Can you fix your project, and then we can try to merge it again.

Thanks!

Valdis

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
120e7c696319: docs/phonemes.md: fix the ust and dst diacritic symbols.

Modified: docs/phonemes.md

#203 myanmar phoneme added

[espeak-ng:master] New Comment on Issue #201 make clean ; make doesn't work.
By ValdisVitolins:

I applied this fix to my branch.

5 New Commits:

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
4fd6ee70c1a5: docs/phonemes.md: describe the tones using properties, not features.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
4209d91271ad: docs/phonemes.md: fix section indenting.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
92f7f0f11ba4: docs/phonemes.md: group the consonant charts.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
d669eba0decf: docs/phonemes.md: combine the pulmonic and non-pulmonic charts into a single chart.

Modified: docs/phonemes.md

[espeak-ng:master] By Valdis Vitolins <valdis.vitolins@...>:
4831ed54064b: Fix for issue #201 Make clean ; make doesn't work

Modified: Makefile.am

[espeak-ng:master] New Comment on Pull Request #203 myanmar phoneme added
By rhdunn:

I have merged the 'make clean' patch. The Myanmar/Burmese changes has issues, so I am not merging that:

1. the file in espeak-ng is ph_burmese, not ph_myanmar (both are valid names for the language);
2. when using ph_myanmar as ph_burmese, I get: ph_burmese(1133): Can't read file: FinalMy/a43.

[espeak-ng/espeak-ng] Pull request closed by rhdunn:

#203 myanmar phoneme added

#203 myanmar phoneme added

Found that updating clean-local target to remove .o/.lo files from src/llibespeak-ng worked:

diff --git a/Makefile.am b/Makefile.am
index 1fb36583..82202bd0 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -55,6 +55,7 @@ install-exec-hook:
 
 clean-local:
    cd src && rm -f *.o *~ && cd ..
 +  cd src/libespeak-ng/ && rm -f *.o *.lo && cd ../..
    cd android && $(GRADLE) clean

Just note that phoneme names from phsource/phonemes file would be: ʒ as Z ð as D or T

I ask if you can send me the audio file for this input: ذَهَبَ الظَبيُ وَ بَقِيَت الضِفدَعَة فِي المُستَنقَعِ ثُمَّ خَرَجَت مِنهُ I can verify if other matters exist within the configuration.

In the description of Sebawayh, [d͡ʒ] is the pronunciation of ج. However, the pronunciation of Arabic had changed. When the linguists have decided to create the Modern Standard Arabic that will replace the Classical Arabic by the 19th Century, they have studied the phonology of the Arabic dialects and replaced Classical pronunciation by the mostly used dialectal arabic one. For example, ج became pronounced as [ʒ]. You can verify this by seeing the main references about this issue that are used by specialists in Computational Linguistics. For example, Al-Aswat al-lughawiyah of Ibrahim Anis and Al-Aswat al-lughawiyah of Muhammad Alkhouli...

Of course, ج is pronounced as [ʒ]. You can listen to it in https://en.wikipedia.org/wiki/FileVoiced_palato-alveolar_sibilant.ogg.

You can view that [dzˤ] is the standard pronunciation, cf. https://en.wikipedia.org/wiki/Arabic_phonology or https://en.wikipedia.org/wiki/Voiced_palato-alveolar_affricate but the [ʒ], is used in dialect, https://en.wikipedia.org/wiki/Arabic_phonology#endnote_6.

you can read

^ The phoneme represented by the Arabic letter ǧīm (ج) has many standard pronunciations: [d͡ʒ] in most of the Arabian Peninsula and as the predominant pronunciation of Literary Arabic, [ɡ] in most of Egypt and some regions in Yemen and Oman. This is also a characteristic of colloquial Egyptian and southern Yemeni dialects.[16] In Morocco and western Algeria, it is pronounced as [ɡ] in some words, especially colloquially. In most north Africa and most of the Levant, the standard is pronounced [ʒ], and in certain regions of the Persian Gulf with [j]. In some Sudanese and Yemeni dialects, it may be either [ɡʲ] or [ɟ] as it used to be in Classical Arabic. Foreign words containing /ɡ/ may be transcribed with ج‎, غ‎, ك‎, ق‎, گ‎, ݣ‎ or ڨ‎, mainly depending on the regional spoken variety of Arabic or the commonly diacriticized Arabic letter . Also, /d͡ʒ~ʒ/ (چ only in Egypt) can be used in loanwords where it isn't the standard pronunciation for the letter ǧīm (ج).

For other phonemes, I will check it.

10 New Commits:

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
073e1653715e: docs/phonemes.md: add the vwl feature to the manner of articulation table.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
2a1309497477: docs/phonemes.md: rename the height and backness sections.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
49a5358db9d2: docs/phonemes.md: move syllability up a level.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
c70465c14dd6: docs/phonemes.md: move consonant release up a level.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
1f600f0684ca: docs/phonemes.md: move frr to the consonant release feature table.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
e69036270695: docs/phonemes.md: move the fortis and lenis, and co-articulation feature tables up a level.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
a5749da4a65f: docs/phonemes.md: use passive instead of other in the place of articulation table.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
488a103c1275: docs/phonemes.md: reorganise the co-articulation features.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
906cbd84659f: docs/phonemes.md: don't list Kirshenbaum features in the remaining feature tables.

Modified: docs/phonemes.md

[espeak-ng:master] By Reece H. Dunn <msclrhd@...>:
9f8c73e41452: docs/phonemes.md: update the introduction text.

Modified: docs/phonemes.md