openai-whisper-api-cli 0.1.1

Simple CLI wrapper for OpenAI Speech-to-Text API

openai-whisper-api-cli

Simple Python CLI wrapper for OpenAI Speech-to-Text.

Features

• Transcribe local audio/video files with OpenAI Speech-to-Text
• API key from --api-key or OPENAI_API_KEY
• Print result to stdout (default)
• Optionally write result to a file
• Supports multiple response formats

Install (Recommended)

python3 -m pip install --user pipx
python3 -m pipx ensurepath
pipx install openai-whisper-api-cli

Verify:

owhisper --version

Upgrade:

pipx upgrade openai-whisper-api-cli

Install From Source

python3 -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"

Authentication

Set env var:

export OPENAI_API_KEY="sk-..."

Or pass key directly:

owhisper transcribe ./audio.mp3 --api-key "sk-..."

--api-key takes precedence over OPENAI_API_KEY.

Usage

Print transcription to stdout:

owhisper transcribe ./audio.mp3

Write transcription to file:

owhisper transcribe ./audio.mp3 --output-file ./transcript.txt

Use custom model and response format:

owhisper transcribe ./audio.mp3 \
  --model gpt-4o-transcribe \
  --response-format json \
  --language en

Supported models

• gpt-4o-transcribe
• gpt-4o-mini-transcribe
• gpt-4o-transcribe-diarize
• whisper-1

Model list source:

• https://developers.openai.com/api/docs/guides/speech-to-text/

CI and Publishing

GitHub Actions workflows:

• .github/workflows/ci.yml: runs tests on push/PR
• .github/workflows/publish.yml: builds and publishes to PyPI on tag push (v*)

PyPI trusted publishing setup

1. Create your project on PyPI: openai-whisper-api-cli.
2. In PyPI project settings, add a Trusted Publisher:
   • Owner: your GitHub org/user
   • Repository: openai-whisper-api-cli
   • Workflow: publish.yml
   • Environment: pypi
3. In GitHub repo settings, ensure environment pypi exists (optional protection rules).

Release

python -m pytest
python -m build
python -m twine check dist/*
git tag v0.1.0
git push origin v0.1.0

Pushing the tag triggers automated publish.

Local manual publish (optional)

python -m build
python -m twine upload dist/*

License

MIT (see LICENSE).