uccharan (Python)

Phonemiser, text normaliser, transliterator and pronunciation engine for text-to-speech and speech recognition: 39 languages (English, Hindi, nineteen more South Asian languages and seventeen European ones), in Rust, Python, JavaScript and C. Apache-2.0, no GPL code or data. (what it does · measurements)

• Phonemes in the shapes Piper, Kokoro, Kitten TTS, Matcha-TTS and sherpa-onnx voices expect, compatible with the espeak-ng output those voices were trained on; an improved mode for voices you train yourself.
• Normalise-only mode: the spoken form of numbers, money, dates and URLs for voices that read text, with spans aligned to the input.
• Transliteration of romanised Hindi, Marathi, Bengali, Tamil, Telugu, Gujarati, Punjabi, Kannada and Malayalam into native script.
• Pronunciation word lists, sound-alike matching of a recogniser's output against your vocabulary, and tools for training voices.

uv add uccharan          # or: uv pip install uccharan
pip install uccharan

import uccharan

uccharan.phonemize("Hello, world.")                          # 'həlˈoʊ, wˈɜːld.'
uccharan.phonemize("नमस्ते दुनिया", lang="hi")                  # 'nəmˈʌsteː dˈʊnɪjˌaː'
uccharan.phonemize("Hi there. Bye!", "en-us", "piper")       # ['hˈaɪ ðˈɛɹ.', 'bˈaɪ!']
uccharan.phonemize("Hi there.", "en-us", "sherpa-onnx")      # [[104, 712, 97, 618, …]]
uccharan.phonemize("I read it yesterday.", improved=True)    # 'aɪ ɹˈɛd ɪt jˈɛstɚdˌeɪ.'
uccharan.phonemize("I read the book.", "en-us", "kokoro")    # 'ˌI ɹˈɛd ðə bˈʊk.'  (misaki conventions, for Kokoro)
uccharan.phonemize_batch(lines, "hi")                        # all cores, GIL released

uccharan.normalize("Rs. 250 on 3/4/2024", "en-us").text      # 'two hundred fifty rupees on March fourth twenty twenty four'
uccharan.normalize("₹5 करोड़ का सौदा 10:30 बजे", "hi").text     # 'पांच करोड़ रुपये का सौदा साढ़े दस बजे'

uccharan.load_pack("xlit-hi")                                # CC BY-SA 4.0 pack, downloaded once
uccharan.transliterate("mujhe kal office jana hai", "hi").text   # 'मुझे कल office जाना है'

names = uccharan.SoundIndex(["Kothari", "Paytm"])
names.apply("send it to Coat Harry", "en-us").text           # 'send it to Kothari'

For code written against phonemizer:

from uccharan.compat import EspeakBackend
backend = EspeakBackend("en-us", preserve_punctuation=True, with_stress=True)
backend.phonemize(["Hello, world."], strip=True)             # ['həlˈoʊ, wˈɜːld.']

A command-line program is installed as uccharan (uccharan -l hi -p piper "नमस्ते").

Guides: normalise-only mode, word lists, sound-alike matching, transliteration, languages.

For training voices and checking pronunciation, uccharan.train and the uccharan-train command label a dataset (training-ready phonemes for Piper, Matcha-TTS and icefall-style recipes, a symbol table and a voice.json), choose a reading script, report phoneme and diphone coverage, and score a recogniser's output against a reference word by word (guide).

English for Kokoro uses a port of misaki with misaki 0.9.4's dictionaries; load_misaki(dir) loads another copy.

The wheel (about 5 MB) carries the data of every language except British English. British English and the Kokoro English data are language packs: downloaded from the GitHub release of the same version the first time they are needed, checked against SHA-256 sums inside the wheel, and cached in ~/.cache/uccharan. On an offline machine put the .ucd files (from the release page) in a directory named by UCCHARAN_DATA_DIR, set UCCHARAN_OFFLINE=1 to forbid downloads, or call uccharan.load_pack(path). The transliteration packs (xlit-hi, …) and the Wiktionary packs (wikipron-en-us, …) are CC BY-SA 4.0 and are fetched only when you call load_pack with their name.

Documentation, measured agreement, speed and end-to-end results: https://github.com/cB-Sigma/uccharan.

Licence

Licence: Apache-2.0. Includes data from the CMU Pronouncing Dictionary (BSD-2-Clause) and misaki (Apache-2.0); see NOTICE. The optional wikipron-* and xlit-* packs are CC BY-SA 4.0; they are separate downloads and are not part of this package.