---
id: kaldifst
version: "1.8.0"
license: Apache-2.0
license_treatment: permissive
maintenance: active
---

# kaldifst

License: permissive · Maintenance: active · Downloads: 295.9K/mo

## What it is and what it does

Kaldifst is a Python wrapper around OpenFst and selected extensions from the Kaldi speech recognition toolkit. It lets you use finite-state transducer operations—such as fstdeterminizestar—directly from Python without needing to install Kaldi itself. The package is compiled to native code and distributed as prebuilt wheels for common platforms and Python versions.

It is primarily used in speech recognition, language modeling, and other NLP tasks where finite-state machines are part of the pipeline. The wrapper bridges the gap between high-level Python code and the lower-level FST algorithms, making it easier to integrate Kaldi-style FST operations into Python applications.

Use it for:

- Build speech recognition pipelines that need FST determinization and composition operations.
- Integrate finite-state transducers into Python-based NLP workflows without a full Kaldi installation.
- Perform language model composition and optimization using OpenFst algorithms from Python.
- Prototype and experiment with FST-based algorithms in interactive environments.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Python wrapper for OpenFst and Kaldi extensions, providing finite-state transducer operations like fstdeterminizestar without requiring Kaldi installation.

Yes, if you need OpenFst or Kaldi FST operations in Python and are on a supported platform (Python 3.10–3.12, macOS/Linux/Windows). The permissive license, active maintenance, zero known vulnerabilities, and prebuilt wheels make it straightforward to adopt. If you require Python 3.8 support, verify source-build feasibility first.

## Install

pip install kaldifst
uv add kaldifst
poetry add kaldifst

## Installing kaldifst

Before you install: Medium install friction due to compiled wheels; prebuilt binaries available for Python 3.10, 3.11, and 3.12 on macOS (Intel and ARM), Linux (x86_64 and aarch64), and Windows. Last release 127 days ago; repository is active with 56 stars.

License in practice: Apache-2.0 permissive license allows commercial and private use with minimal restrictions; suitable for most projects.

Quickstart:

pip install kaldifst
import kaldifst
# Use OpenFst operations (e.g., fstdeterminizestar) via the wrapper

Requires Python >=3.8; prebuilt wheels available for Python 3.10, 3.11, 3.12 on select platforms.

Verify before relying:

- Specific API surface and which Kaldi FST extensions are exposed beyond fstdeterminizestar.
- Whether the package supports Python 3.8 via source build or only via prebuilt wheels for 3.10–3.12.
- Performance characteristics and typical use patterns in speech recognition or NLP pipelines.

## Package facts

- License: Apache-2.0 (permissive)
- Python support: supports_current
- Install friction: medium
- Maintenance: active
- Downloads: 295.9K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

finite state transducer python, openfst python wrapper, kaldi fst operations, speech recognition fst, fstdeterminizestar python, speech-recognition, finite-state-machines, nlp

[View on SkillFed](https://skillfed.io/packages/kaldifst) · [View on PyPI](https://pypi.org/project/kaldifst/)