yup,
its not there, nor in the setup.py
Just a heads up
thanks for the hard work here
a.

Hi Michael,

I've been experimenting with gruut-ipa and Zamia lexicon.txt from the popular models 'kaldi-generic-de-tdnn_250-r20190328' and ''kaldi-generic-en-tdnn_250-r20190609" and I'm having trouble getting the expected results.
As far as I understand Zamias lexicon.txt: is in sampa format, so I selected two words from the German file:

hallo --> h '{ l @ U
welt --> v 'E l t

and then used espeak-ng to generate phonems:

# espeak IPA phonems:
espeak-ng -v de -x -q --sep=" " --ipa "hallo"
h ˈa l oː
espeak-ng -v de -x -q --sep=" " --ipa "welt"
v ˈɛ l t

# espeak default phonems:
h 'a l o:
v 'E l t

Finally I've tried to convert the espeak results to sampa with gruut-ipa:

python3 -m gruut_ipa convert ipa sampa "h ˈa l oː"
h "a l o:

python3 -m gruut_ipa convert espeak sampa "h 'a l o:"
h "a 5 o:

python3 -m gruut_ipa convert ipa sampa "v ˈɛ l t"
v "E l t

python3 -m gruut_ipa convert espeak sampa "v 'E l t"
v "E 5 t

But none of the results matches the lexicon.txt entries.
Any help or hints would be appreciated! :-)

I'm currently packaging larynx + deps for Arch Linux and I encountered an issue with gruut-ipa: setuptools includes the tests in the package, which does not play well with Arch Linux packaging - see rhasspy/phonemes2ids#1 for details.

I propose the following patch:

--- a/setup.py  2021-11-11 20:26:11.000000000 +0100                                                                                                                                                               
+++ b/setup2.py 2022-01-11 01:13:56.813670352 +0100
@@ -36,7 +36,7 @@
     author="Michael Hansen",
     author_email="[email protected]",
     url="https://github.com/rhasspy/gruut-ipa",
-    packages=setuptools.find_packages(),
+    packages=setuptools.find_packages(exclude=['tests']),
     package_data={"gruut_ipa": data_files + ["py.typed"]},
     install_requires=requirements,
     extras_require={':python_version<"3.7"': "dataclasses"},

There is an issue when we convert r\ from sampa to ipa. I'd expect it to be ɹ, but we get ɰ̩ instead.
The issues is that when we construct SAMPA_TO_IPA dictionary, we reverse the IPA_TO_SAMPA one, but we have r\ as a value twice in it, so it get's rewritten.

Hello, I'm trying to use gruut-ipa 0.10.0 on Windows 10 and Python 3.9.
I need to group phones into phonemes for English:

$ python3 -m gruut_ipa phonemes en-us "/dʒʌst ə kaʊ/"
d͡ʒ ʌ s t ə k aʊ

How would this look in the python interpreter? I've tried

>>> import gruut_ipa
>>> help(gruut_ipa.Phonemes)

But I couldn't find an answer there.

Hi,

I recently stumbled on this project and I have to say I'm quite impressed. It's a fresh take on all the other projects related to IPA in Python. Keep it up!

One thing I noticed while playing around was that in the consonant descriptions, we don't see the manner of articulation, neither from the command line nor when calling it from code. For example:

## command line
python -m gruut_ipa describe 'f' | jq .
{
  "text": "f",
  "letters": "f",
  "tone": "",
  "tones": [],
  "accents": [],
  "stress": "",
  "type": "Consonant",
  "place": "labio-dental",
  "voiced": false,
  "nasalated": [],
  "raised": [],
  "elongated": false
}

## from python script
gruut_ipa.Phoneme('f').to_dict()
{'text': 'f',
 'letters': 'f',
 'tone': '',
 'tones': [],
 'accents': [],
 'stress': '',
 'type': 'Consonant',
 'place': 'labio-dental',
 'voiced': False,
 'nasalated': [],
 'raised': [],
 'elongated': False}

I can tell the information is there if I call the consonant object, it just seems like the information is lost when converting this object to the dict/json output.

gruut_ipa.Phoneme('f').consonant
Consonant(ipa='f', type=<ConsonantType.FRICATIVE: 'fricative'>, place=<ConsonantPlace.LABIO_DENTAL: 'labio-dental'>, voiced=False, velarized=False, sounds_like=<ConsonantSoundsLike.NONE: ''>, length=<PhonemeLength.NORMAL: 'normal'>, alias_of=None)

I think it has to do with the code in the main file calling for the 'type' of the consonant object, but I'm not in a position to debug it at the moment. The internal type is fricative and the python type is Consonant. This could be fixed by example by referring to the consonant 'type' as the consonant 'manner', which is in line with IPA standards.

Looking forward to your feedback. Thanks again for the awesome tool!

Hi,

I'm coming in from using coqui, but I realize this is a gruut issue more than a coqui issue, so here's my question -

For the following text - 'In the news today - Afghanistan, COVID 19, Corona Virus, Australia, India, and the USA.'

I ran the following command and got the below phonemes -

> echo 'In the news today - Afghanistan, COVID 19, Corona Virus, Australia, India, and the USA.' | python3 -m gruut en-us tokenize | python3 -m gruut en-us phonemize | jq -c ".pronunciation_text"
"ˈɪ n ð ə n j ˈu z t ə d ˈeɪ æ f ɡ ˈæ n ɪ s t ˌæ n | k ˈoʊ v ɪ d n ˈaɪ n t ˈi n | k ɚ ˈoʊ n ə v ˈaɪ ɹ ə s | ɔ s t ɹ ˈeɪ l j ə | ˈɪ n d i ə | ˈæ n d ð ə ˌʌ s ə ‖"

From what I can see, the translation for "USA" is wrong and for corona virus is iffy, at best. Coqui ends up speaking "USA" as "you-sa" (with [sa] as in [sardonic]) instead of individual letters "U", "S", "A".

How can I fix this? Is there some text cleanup I can do? I tried passing "U S A" with spaces but that doesn't fare so well either.

I face the same problem with saying "AWS", because it translates it as "ɔ z".

I can see that this is based on the phonemes text file here - https://github.com/rhasspy/gruut-ipa/blob/master/gruut_ipa/data/en-us/phonemes.txt

ɔ l[aw] ɔː ɒ

Again, splitting it up doesn't help. Could you tell me if I can do something to get gruut to acknowledge that these are individual letters that need to be spoken out exactly as they are written?

I'm trying to use your awesome library for some language investigation, and faced with a problem with palatalized consonants.

Your library thinks about ʲ (\u02B2, IPA number 421) as about separate phone. But as far as I understand it should be thought as diacritic symbol that marks base consonant as palatalized one. So correct output for
>>> Pronunciation.from_string('konʲ').phones
is not
[k, o, n, ʲ]
but
[k, o, nʲ]

Kindly let me know if my assumption is correct (I'm not a linguist, and probably understand IPA in a wrong way), and I'll make you a pull request to fix that.

Hi, thanks for working on everything Rhasspy, much appreciated.

If you could use Git tags on this repository to publish new versions we could better reference and update this package in downstream Linux distros.

Thanks for considering!

gruut-ipa phonemes de-de "lecken"
gives "l eː ç k eː n", however, "ck" is one sound "k" here.

gruut-ipa phonemes de-de "Straße"
gives "S t ʁ a ß e", however, "a" in "Straße" is long (i.e. "a:").

From my understanding there should be at least one phone for each phoneme in English of which there are supposed ot be 44 phonemes so presumably at least 44 phones. When I do gruut-ipa print --language en-gb I get 33 phones (and 36 for en-us)? The phoneme data files have 42 and 40 phonemes respectively?

Also, the phones command doesn't seem to handle dipthong vowels at all, splitting them up into two phones?

What am I missing here?