V2kAudioProcessor

Process audio to find voices in it. Used to process V2k recordings and find the audio within it.

Batch denoising CLI

You can quickly run the spectral denoising pipeline over one or more audio files (or folders full of files) without writing code:

python -m audio.pipeline path/to/input.wav another_dir/ \
  --output-dir denoised_outputs \
  --preset aggressive \
  --multiband-gate --wiener

• Accepts individual files and directories (searched recursively with --glob, default *.wav).
• Preserves relative paths inside the output directory and appends a _denoised_<preset>.wav suffix.
• Supports the same optional stages as the library API: multi-band gating, Wiener filtering, and spectral subtraction.

The underlying functions are available from Python if you want to script your own evaluation against a set of test clips:

from pathlib import Path
from audio import process_batch

outputs = process_batch([Path("tests/inputs")], Path("tests/outputs"), preset="conservative")
for out in outputs:
    print(out)
## Overview

The processing pipeline loads an input file, optionally applies a manual gain
boost, runs a placeholder denoiser, and can then normalize loudness and/or add a
band-pass filter before saving the result. Loading prefers `soundfile` and
falls back to `ffmpeg` for uncommon codecs. Audio is resampled to a fixed sample
rate (default 16 kHz) and converted to mono for consistent downstream
processing.

## Command line interface

Run the CLI with:

python -m src.cli INPUT OUTPUT [--target-sr 16000] [--gain-db 3.0] [--normalize] [--bandpass]

Available options are also listed in `--help` output:

* `--target-sr` – resample input to this rate (defaults to 16 kHz).
* `--gain-db` – apply a manual preamp (in decibels) before denoising.
* `--normalize` – normalize loudness to the EBU R128 target (-23 LUFS).
* `--bandpass` – apply an 80 Hz–8 kHz Butterworth band-pass filter.

Example:

python -m src.cli noisy.wav cleaned.wav --gain-db 6 --normalize --bandpass