A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.
Python artificial-intelligencedagdata-sciencedata-visualizationdataflowdeveloper-toolsmachine-learningnotebookspipelinepythonreactivesqlweb-app
# /// script
# requires-python = ">=3.12"
# dependencies = ["marimo>=0.25.0"]
#
# [tool.pixi.workspace]
# channels = ["conda-forge"]
#
# [tool.pixi.dependencies]
# ffmpeg = "*"
# ///
import marimo
__generated_with = "0.25.2"
app = marimo.App()
@app.cell(hide_code=True)
def _(mo):
mo.md("""
With [Pixi installed](https://pixi.prefix.dev/latest/installation/), run:
```bash
marimo edit --sandbox=pixi examples/misc/pixi_ffmpeg.py
```
""")
@app.cell
def _():
import marimo as mo
pitch = mo.ui.slider(220, 880, step=20, value=440, label="Pitch (Hz)")
pitch
return mo, pitch
@app.cell
def _(mo, pitch):
import io
import subprocess
tone = subprocess.run(
[
"ffmpeg",
"-f",
"lavfi",
"-i",
f"sine=frequency={pitch.value}:duration=2",
"-f",
"wav",
"pipe:1",
],
capture_output=True,
check=True,
)
mo.audio(io.BytesIO(tone.stdout))
if __name__ == "__main__":
app.run()