Convolution Reverb

A Python script that applies convolution reverb to audio files using real impulse response recordings. It convolves a dry input signal with an impulse response (IR) captured from a physical space, producing a realistic reverberation effect. The script outputs both a fully wet (100% reverb) file and a blended wet/dry mix.

How It Works

Convolution reverb works by mathematically combining a dry audio signal with an impulse response -- a recording of a short, sharp sound (such as a clap) in a real acoustic environment. The result is an audio file that sounds as if it were recorded in that space.

The script performs the following steps:

1. Loads the input audio and impulse response files.
2. Converts stereo files to mono (if necessary).
3. Convolves the two signals using scipy.signal.convolve.
4. Normalizes the output to 90% peak level to avoid clipping.
5. Creates a wet/dry blend of the convolved signal and the original.
6. Writes both results to disk as .wav files.

Requirements

• Python 3
• NumPy
• SciPy
• SoundFile

Install dependencies with:

pip install numpy scipy soundfile

Usage

Place your input audio file and impulse response file in the same directory as the script, then run:

python convolve.py

The script uses hardcoded file paths and parameters. To change them, edit convolve.py directly:

Line	Variable	Default	Description
6	Input file	input.wav	Path to the dry audio file
7	Impulse response file	trapphus.wav	Path to the impulse response file
25	Normalization ceiling	0.9	Peak normalization target (0.9 = ~0.9 dB headroom)
32	wet_level	0.5	Reverb level in the blended output (0.0 to 1.0)
33	dry_level	1 - wet_level	Dry signal level in the blended output

Output

The script produces two files:

• output.wav -- The fully wet convolved signal (100% reverb).
• wet_and_dry_blend.wav -- A mix of the dry and wet signals, controlled by the wet_level parameter.

Included Impulse Responses

Three impulse response recordings are included, all captured at 48 kHz / 24-bit:

File	Description
kyrka.wav	A church
eva_von_bahr.wav	Eva von Bahr lecture hall
trapphus.wav	A stairwell

To switch between them, change the impulse response filename on line 7 of convolve.py:

impulse_response, ir_sr = sf.read('kyrka.wav')