Skip to main content

Automatic accentuation for texts in Russian

Accentuation is a common task in such speech-related fields as text-to-speech, speech recognition and language learning. This library is used to mark stressed vowels in given texts using the data from Wiktionary and syntactic analysis of Spacy.

Installation

Python 3.10 and above supported.

pip install tsnorm

General usage

from tsnorm import Normalizer


normalizer = Normalizer(stress_mark=chr(0x301), stress_mark_pos="after")
normalizer("Словно куклой в час ночной теперь он может управлять тобой")

# Output: Сло́вно ку́клой в час ночно́й тепе́рь он мо́жет управля́ть тобо́й

Change stress mark and its position

normalizer = Normalizer(stress_mark="+", stress_mark_pos="before")
normalizer("Трупы оживали, землю разрывали")

# Output: Тр+упы ожив+али, з+емлю разрыв+али

Stress yo (Ё)

normalizer = Normalizer(stress_mark="+", stress_mark_pos="before", stress_yo=True)
normalizer("Погаснет день, луна проснётся, и снова зверь во мне очнётся")

# Output: Пог+аснет день, лун+а просн+ётся, и сн+ова зверь во мне очн+ётся

Stress monosyllabic words

normalizer = Normalizer(stress_mark="+", stress_mark_pos="before", stress_monosyllabic=True)
normalizer("Панки грязи не боятся, кто устал — ушёл сдаваться!")

# Output: П+анки гр+язи н+е бо+ятся, кт+о уст+ал — ушёл сдав+аться!

Change minimum length of words to be stressed

normalizer = Normalizer(stress_mark="+", stress_mark_pos="before", stress_monosyllabic=True)
normalizer("Разбежавшись, прыгну со скалы, вот я был и вот меня не стало")

# Output: Разбеж+авшись, пр+ыгну с+о скал+ы, в+от +я б+ыл +и в+от мен+я н+е ст+ало


normalizer = Normalizer(stress_mark="+", stress_mark_pos="before", stress_monosyllabic=True, min_word_len=2)
normalizer("Разбежавшись, прыгну со скалы, вот я был и вот меня не стало")

# Output: Разбеж+авшись, пр+ыгну с+о скал+ы, в+от я б+ыл и в+от мен+я н+е ст+ало


normalizer = Normalizer(stress_mark="+", stress_mark_pos="before", stress_monosyllabic=True, min_word_len=3)
normalizer("Разбежавшись, прыгну со скалы, вот я был и вот меня не стало")

# Output: Разбеж+авшись, пр+ыгну со скал+ы, в+от я б+ыл и в+от мен+я не ст+ало

Expand normalizer dictionary

from tsnorm import Normalizer, CustomDictionary, WordForm, Lemma, WordFormTags, LemmaPOS


normalizer = Normalizer("+", "before")

normalizer("Охотник Себастьян, что спал на чердаке")
# Output: Ох+отник Себастьян, что спал на чердак+е

dictionary = CustomDictionary(
    word_forms=[
        WordForm("Себастьян", 7, WordFormTags(singular=True, nominative=True), "Себастьян")
    ],
    lemmas=[
        Lemma("Себастьян", LemmaPOS(PNOUN=True))
    ]
)

normalizer.update_dictionary(dictionary)

normalizer("Охотник Себастьян, что спал на чердаке")
# Output: Ох+отник Себасть+ян, что спал на чердак+е

It's also possible to pass CustomDictionary at normalizer initialization:

normalizer = Normalizer("+", "before", custom_dictionary=dictionary)

To add your custom words to normalizer dictionary you must pass two lists to CustomDictionary:

  1. a list of WordForm objects, which are forms of each word with case, tense and lemma information, as well as the positions of stressed letters,
  2. a list of Lemma objects, which are records of lemmas with their parts of speech.

Parts of speech for lemmas are configured using the LemmaPOS class which stores universal POS tags.

Acknowledgement

This library is based on code by @einhornus from his article on Habr.

Metadata

Release files for tsnorm 1.1.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distribution (wheel)

Table of built distributions (wheels) for tsnorm 1.1.2
File Interpreter ABI Platform
tsnorm-1.1.2-py3-none-any.whl Python 3 none any Details

Release files / tsnorm-1.1.2-py3-none-any.whl

Download URL tsnorm-1.1.2-py3-none-any.whl
Size 17.2 MB
Tags Python 3
SHA-256 checksum
How to use checksums
9cffa03a38f3382a362c4d14f2b0810b0feed547a8deceb3fea0f292218928d9
BLAKE2b-256 checksum
How to use checksums
d699dec2a2b0666e93631b2365974ed5b9bdfb6feab43929462a073b5d742fd1
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.0.1 CPython/3.9.21

Release history Release notifications | RSS feed

This release

1.1.2 This release

1 release file

1.1.0

1 release file

1.0.0

1 release file

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page