g2p-en

What it is and what it does

g2p_en is a Python module that converts English text to phoneme sequences, a critical preprocessing step for speech synthesis and text-to-speech systems. It handles the irregularity of English pronunciation by combining three strategies: dictionary lookup via the CMU Pronouncing Dictionary, part-of-speech-based disambiguation for homographs (words with multiple pronunciations), and neural sequence-to-sequence prediction for out-of-vocabulary words. The module also normalizes numbers and abbreviations into spelled-out forms before conversion.

The package depends on numpy for inference (TensorFlow was removed to avoid GPU requirements and API churn), nltk for part-of-speech tagging and dictionary access, inflect for morphological operations, and distance for string matching. It is designed for Python 3.x and installs as a pure wheel with low friction. However, the project has been abandoned since late 2019, with no maintenance or updates since early 2023, so users should be prepared to handle any compatibility issues with modern dependency versions independently.

Use it for:

• Preprocessing text for text-to-speech (TTS) systems where accurate phoneme sequences are required for synthesis quality.
• Building speech recognition training data pipelines that need ground-truth phoneme alignments from orthographic text.
• Linguistic analysis of English pronunciation patterns, homograph disambiguation, and out-of-vocabulary word handling.
• Augmenting speech datasets with phonetic transcriptions for acoustic model training.
• Educational tools that teach English pronunciation by mapping written words to their phonetic representations.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Converts English text to phoneme sequences using dictionary lookup, part-of-speech disambiguation, and neural prediction for out-of-vocabulary words.

Yes, if you need English grapheme-to-phoneme conversion and can tolerate an abandoned package. The module is stable, has no known vulnerabilities, and works well for its narrow purpose. Install it only if you are comfortable maintaining compatibility with its dependencies yourself and do not expect upstream fixes or feature updates. For active projects requiring long-term support, consider whether a maintained alternative exists.

Install

g2p-en on PyPI

pip install g2p-en

uv add g2p-en

poetry add g2p-en

Installing g2p-en

Before you install

Low install friction with a pure-Python wheel and four straightforward dependencies. However, the package is abandoned—last release was 2019-12-31 and last commit 2023-01-05—so expect no maintenance, bug fixes, or updates to handle modern dependency versions.

License in practice

Apache Software License (permissive) allows free use, modification, and distribution with minimal restrictions, making it safe to integrate into most projects without licensing concerns.

Quickstart

pip install g2p_en

from g2p_en import G2p

g2p = G2p()
output = g2p("I have $250 in my pocket.")
print(output)

nltk data files (averaged_perceptron_tagger, cmudict) are downloaded automatically on first run; no explicit setup required but network access is needed during initialization.

Verify before relying

• Whether the package's neural model remains accurate for modern English or newly coined words beyond its training data.
• Compatibility with current versions of numpy, nltk, inflect, and distance given the package's abandoned status.
• Performance characteristics (latency, memory) for batch processing or real-time speech synthesis pipelines.

Package facts

Evidence: g2p_en-2.1.0-py3-none-any.whl

Keywords: g2p, g2p_en, g2pE

Tags

More Linguistic packages