Minimal Example#

VoxAtlas is organized around modular feature extractors and a pipeline that coordinates them.

This minimal example shows the intended workflow at a high level:

  1. Load or prepare your conversational data.

  2. Choose the feature extractors you want to run.

  3. Execute the pipeline.

  4. Inspect or export the resulting feature outputs.

Example outline#

The snippet below is fully runnable and uses the built-in acoustic.pitch.dummy extractor to verify the pipeline path end-to-end.

import numpy as np

from voxatlas.audio.audio import Audio
from voxatlas.pipeline import Pipeline

audio = Audio(waveform=np.zeros(16000, dtype=np.float32), sample_rate=16000)
config = {
    "features": ["acoustic.pitch.dummy"],
    "pipeline": {"n_jobs": 1, "cache": False},
}

results = Pipeline(audio=audio, units=None, config=config).run()
output = results.get("acoustic.pitch.dummy")

print(output.feature, output.unit, float(output.values.iloc[0]))

To run it from a repository checkout without installing the package, execute this from the repository root:

PYTHONPATH=src python - <<'PY'
import numpy as np

from voxatlas.audio.audio import Audio
from voxatlas.pipeline import Pipeline

audio = Audio(waveform=np.zeros(16000, dtype=np.float32), sample_rate=16000)
config = {"features": ["acoustic.pitch.dummy"], "pipeline": {"n_jobs": 1, "cache": False}}

results = Pipeline(audio=audio, units=None, config=config).run()
output = results.get("acoustic.pitch.dummy")
print(output.feature, output.unit, float(output.values.iloc[0]))
PY

Where to go next#