skillfed

kaldi-python-io

A pure python IO interface for data accessing in kaldi

kaldi-python-io v1.2.2 579.0K downloads/30d#5,921 on PyPI39
Permissive license Apache V2.0 Abandoned released

What it is and what it does

kaldi-python-io is a Python wrapper that provides access to Kaldi's native data formats—binary archives (.ark), script files (.scp), and neural network training examples (.egs)—without requiring direct interaction with Kaldi's command-line tools. It offers reader and writer classes for sequential and random-access patterns, making it useful for machine learning pipelines that depend on Kaldi's feature extraction or alignment output.

The package has been abandoned since March 2021 and receives no maintenance. It requires Python 3.6 or later and depends on an external Kaldi installation. The lack of runtime dependencies in PyPI suggests it wraps compiled Kaldi binaries or relies on system-level configuration, which may complicate setup and limit compatibility with newer Kaldi versions.

Use it for:

  • Load pre-computed Kaldi features or alignments into Python for downstream machine learning model training
  • Write feature matrices to Kaldi archive format for integration with Kaldi-based speech processing pipelines
  • Read Kaldi nnet3 training examples (.egs) for neural network training without re-implementing Kaldi's serialization
  • Iterate over large Kaldi archives sequentially to avoid loading entire datasets into memory
  • Access random entries in Kaldi script files for validation or debugging speech recognition experiments

Worth the install?

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

Reads and writes Kaldi binary archives, alignment files, and neural network training examples in Python, providing sequential and random-access interfaces to Kaldi's data formats.

No. The package is abandoned (no updates since 2021), has high install friction due to undocumented Kaldi system dependencies, and offers no clear path to working with current Kaldi or Python versions. Unless you are maintaining legacy code that already depends on it, consider using Kaldi's native Python bindings or alternative speech-processing libraries with active maintenance.

Install

kaldi-python-io on PyPI

pip

pip install kaldi-python-io

uv

uv add kaldi-python-io

poetry

poetry add kaldi-python-io

Installing kaldi-python-io

Before you install

High install friction: the package is abandoned (last release 2021-03-18, no activity for 1975 days) and has no runtime dependencies listed, suggesting it may require external Kaldi libraries or system configuration not captured in PyPI metadata.

License in practice

Licensed under Apache V2.0 (permissive), allowing commercial and private use with minimal restrictions, though the abandoned status means no ongoing legal or security maintenance.

Quickstart

pip install kaldi-python-io

from kaldi_python_io import ArchiveReader
ark_reader = ArchiveReader("copy-feats ark:foo.ark ark:- |")
for key, value in ark_reader:
    print(key)

Requires Kaldi to be installed and configured on the system; the package is a Python wrapper around Kaldi binaries, not a standalone library.

Verify before relying

  • Whether Kaldi system libraries must be installed separately and how to configure them
  • Whether the package works with current Kaldi versions or only legacy releases
  • Actual runtime dependencies not reflected in PyPI metadata

Package facts

License Apache V2.0 (permissive)
Python support supports the current Python release (>=3.6)
Install friction high — source build required
Runtime dependencies none
Maintenance abandoned — 1,975 days since the last release
Last repo commit
First released
Downloads 578,980/month — #5,921 on PyPI (30-day window, as of 2026-08-14)
Known vulnerabilities none known (OSV.dev, checked 2026-08-14)

Evidence: kaldi-python-io-1.2.2.tar.gz

Tags

kaldi data io pythonread kaldi ark fileskaldi archive readerkaldi alignment readerkaldi nnet3 exampleskaldi scp readerkaldi feature extraction
kaldi-integrationaudio-processingabandoned

More Artificial Intelligence packages