phonemizer

What it is and what it does

Phonemizer is a Python wrapper around multiple text-to-speech and phoneme-extraction backends that converts written words and sentences into their phonetic representations. It supports four backends—espeak-ng (IPA output, 100+ languages), espeak-mbrola (SAMPA, 35 languages), festival (US English, syllable-level tokenization), and segments (user-defined grapheme-to-phoneme mappings)—each with different speed, language coverage, and output format trade-offs. You choose which backend to use based on your language, required phoneme alphabet, and whether you need syllable-level or word-level boundaries.

The package provides both a command-line tool (`phonemize`) and a Python API (`phonemizer.phonemize`). It is actively maintained, has low install friction (pure Python with minimal dependencies), and is published under GPLv3+. The main constraint is that it requires at least one external backend system library to be installed separately; the package itself is a Python interface to those tools.

Use it for:

• Generate IPA transcriptions for linguistic research or speech corpus annotation across many languages.
• Prepare phonetic training data for automatic speech recognition (ASR) or text-to-speech (TTS) systems.
• Extract syllable-level phonetic boundaries for prosody analysis or speech synthesis in festival-supported contexts.
• Build custom phonemization pipelines using user-defined grapheme-to-phoneme mappings via the segments backend.
• Batch-process large text corpora into phonetic form for phonological or acoustic studies.
• Integrate phonetic transcription into NLP pipelines for multilingual phonetic feature extraction.

Worth the install?

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

Phonemizer converts written text into phonetic representations (IPA, SAMPA, or custom phoneme sets) across many languages using pluggable backends like espeak, festival, and segments.

Yes. Phonemizer is a well-maintained, actively developed tool with low install friction and no known vulnerabilities. It fills a clear niche in phonetic text processing across many languages. The GPLv3+ license is a constraint only if you need to build proprietary closed-source software; for research, open-source, and academic use, it is unencumbered. The main gotcha is that you must install one of the external backends separately, but that is by design and well-documented.

Install

phonemizer on PyPI

pip install phonemizer

uv add phonemizer

poetry add phonemizer

Installing phonemizer

Before you install

Low install friction: pure Python wheel with four lightweight runtime dependencies (joblib, attrs, dlinfo, typing-extensions). Active maintenance with a release 14 days ago and 1567 repository stars.

License in practice

GPLv3+ copyleft license requires that any derivative work or distribution must also be licensed under GPLv3 or later. Suitable for open-source projects but incompatible with proprietary closed-source applications.

Quickstart

pip install phonemizer
from phonemizer.phonemize import phonemize
result = phonemize('hello', language='en-us', backend='espeak')

Requires one of the external TTS/phonemization backends (espeak-ng, festival, or segments) to be installed on the system; the Python package alone is a wrapper.

Verify before relying

• Whether backend installation (espeak-ng, festival, segments) is automatic or requires manual system setup on each platform.
• Performance characteristics and memory usage for large-scale batch phonemization tasks.
• Accuracy and coverage differences between the four backends for specific language pairs.

Package facts

Evidence: phonemizer-3.4.0-py3-none-any.whl

Keywords: linguistics, G2P, phone, espeak, festival, TTS

Tags

More Linguistic packages