pyworld
PyWorld: a Python wrapper for WORLD vocoder
Decision gist · record as of 2026-08-14
Yes, if you need speech vocoding and can tolerate compilation friction. PyWorld is stable, has no known vulnerabilities, and is well-suited for speech analysis and synthesis tasks. However, verify the license terms first (they are not recorded in package metadata), and be aware that the project is dormant—expect no active maintenance. Windows users will have the easiest installation experience.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- WORLD vocoder is designed for speech sampled at ≥16 kHz; applying it to lower sample rates will fail.
- Compilation from source requires a C++ build toolchain.
- Medium friction: requires compilation against the bundled WORLD C++ library.
License · maintenance · safety
(unclear) — License status is unclear—no SPDX identifier or raw license text is recorded in the package metadata. Users should verify the license terms in the repository before adopting this package in commercial or restricted contexts.
last release 2025-01-20 (571 days) · last repo commit 2025-01-21 · 790 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 209,434 downloads/mo, #9,513 on PyPI
Alternatives
Verify before relying
import pyworld as pw
import numpy as np
# Assuming x is a numpy array of audio samples and fs is sample rate
f0, t = pw.dio(x, fs)
f0 = pw.stonemask(x, f0, t, fs)
sp = pw.cheaptrick(x, f0, t, fs)
ap = pw.d4c(x, f0, t, fs)
y = pw.synthesize(f0, sp, ap, fs)- Whether the license is compatible with your intended use (metadata does not specify).
- Whether the package builds successfully on macOS or non-Windows Linux distributions without additional system dependencies beyond those documented.
- Specific Cython version requirements for building from source on different platforms.
What it is and what it does
PyWorld is a Python wrapper around the WORLD vocoder, a fast speech analysis and synthesis tool. It decomposes audio into three acoustic features: f0 (pitch contour), sp (harmonic spectral envelope), and ap (aperiodicity), which can then be modified and resynthesized. The package provides low-level functions (dio, stonemask, cheaptrick, d4c) for each analysis step, plus a convenience function (wav2world) that applies all steps with default parameters.
The package depends only on numpy and wraps compiled C++ code, so installation involves building against the bundled WORLD library. Prebuilt wheels are available for Windows and several Python versions, but Linux and macOS users typically need to compile. The project is dormant (last commit 2025-01-21, no recent activity), though it remains functional and has no known vulnerabilities.
Use it for
- Extract pitch and spectral features from speech for analysis, visualization, or machine-learning feature engineering.
- Modify pitch or spectral properties of speech and resynthesize it for voice conversion or prosody manipulation.
- Analyze speech quality or characteristics in audio processing research or speech enhancement pipelines.
- Generate synthetic speech variants by decomposing, altering, and resynthesizing audio parameters.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you need speech vocoding and can tolerate compilation friction.
PyWorld is stable, has no known vulnerabilities, and is well-suited for speech analysis and synthesis tasks. However, verify the license terms first (they are not recorded in package metadata), and be aware that the project is dormant—expect no active maintenance. Windows users will have the easiest installation experience.
Install
pyworld on PyPI
Before you install
Medium friction: requires compilation against the bundled WORLD C++ library. Prebuilt wheels exist for Windows and several Python versions, but Linux and macOS users may need to build from source. Last release was recent (2025-01-20), though the project is dormant overall.
WORLD vocoder is designed for speech sampled at ≥16 kHz; applying it to lower sample rates will fail. Compilation from source requires a C++ build toolchain.
License in practice
License status is unclear—no SPDX identifier or raw license text is recorded in the package metadata. Users should verify the license terms in the repository before adopting this package in commercial or restricted contexts.
Quickstart
import pyworld as pw
import numpy as np
# Assuming x is a numpy array of audio samples and fs is sample rate
f0, t = pw.dio(x, fs)
f0 = pw.stonemask(x, f0, t, fs)
sp = pw.cheaptrick(x, f0, t, fs)
ap = pw.d4c(x, f0, t, fs)
y = pw.synthesize(f0, sp, ap, fs)Verify before relying
- Whether the license is compatible with your intended use (metadata does not specify).
- Whether the package builds successfully on macOS or non-Windows Linux distributions without additional system dependencies beyond those documented.
- Specific Cython version requirements for building from source on different platforms.
Package facts
| License | Not declared unclear |
| Python support | Not specified |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | 1 packagenumpy |
| Maintenance | Dormant 571 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 209,434 / month, #9,513 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
Evidence: pyworld-0.3.5-cp310-cp310-win_amd64.whl; pyworld-0.3.5-cp311-cp311-win_amd64.whl; pyworld-0.3.5-cp312-cp312-win_amd64.whl; pyworld-0.3.5-cp313-cp313-win_amd64.whl; pyworld-0.3.5-cp36-cp36m-win32.whl; pyworld-0.3.5-cp36-cp36m-win_amd64.whl; pyworld-0.3.5-cp37-cp37m-win32.whl; pyworld-0.3.5-cp37-cp37m-win_amd64.whl; pyworld-0.3.5-cp38-cp38-win_amd64.whl; pyworld-0.3.5-cp39-cp39-win_amd64.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “speech vocoder pitch extraction”
- pyworldPyWorld wraps the WORLD vocoder to decompose speech audio into pitch,…
- torchfcpeTorchFCPE estimates fundamental frequency (pitch) from audio using a…
- torchcrepePyTorch implementation of the CREPE pitch tracker that estimates…
Give your agent the search over MCP, or paste the wish link into any chat.
More Scientific/Engineering packages
NumPy provides an N-dimensional array object and a comprehensive suite of mathematical, linear algebra, Fourier transform, and random number functions for scientific computing in Python.
pandas provides fast, flexible data structures (Series and DataFrame) for loading, cleaning, transforming, and analyzing labeled or relational data in Python.
scipy provides numerical algorithms for mathematics, science, and engineering—including optimization, integration, linear algebra, Fourier transforms, signal and image processing, and ODE solvers—built on numpy arrays.
scikit-learn provides a comprehensive Python library for supervised and unsupervised machine learning, including classification, regression, clustering, dimensionality reduction, and model evaluation tools built on NumPy and SciPy.
Install it if you need to train, evaluate, or deploy supervised or unsupervised learning models.
dill extends Python's pickle module to serialize and deserialize a much wider range of Python objects, including functions, lambdas, classes, and interpreter sessions, to byte streams for storage or network transmission.
Multiprocess is an enhanced fork of Python's standard multiprocessing library that uses dill for better serialization, allowing you to spawn processes with a threading-like API and share complex objects between them.
Install it if you use multiprocessing and encounter pickle serialization limits with lambdas or complex objects.
See also vocos · torchfcpe · torchcrepe · voxcpm · python_speech_features · chatterbox-tts · stftpitchshift · samplerate · aubio · python-stretch