Python Swept Sine

A python implementation of the synchronised swept sine based on Novak et al. 2015 and Farina 2000.

Getting started

pip install -e .

import sounddevice as sd

from sweptsine import SweptSine

my_swept_sine = SweptSine(
    fs=48000, f1=20, f2=20000, duration=10
    scale_dBFS=-6, # eg, to adjust the sweep amplitude at source
    fade_in=0.5, fade_out=0.5, fade_shape="cosine", # eg, to reduce ringing
    pad_start=1, pad_end=10 # eg, to account for latency and decay time
)

measurement = sd.playrec(my_swept_sine.sweep, my_swept_sine.fs, channels=1, blocking=True)

impulse_response = my_swept_sine.deconvolve(measurement)

During the init the sweep parameters are calculated and the sweep (my_swept_sine.sweep) and inverse filter (my_swept_sine.inverse) signals are generated. This includes calculating the actual sweep duration (my_swept_sine.actual_duration) required for phase synchronicity (see Novak et al. 2015, eq. 48).

Measuring using .wav (for example in a DAW)

Signal generation and analysis can be separated by working with .wav files. For example if the measurement recording is done in a DAW.

my_swept_sine.save_sweep_as_wav(path="sweeps", prefix="my_sweep_name")

# creates "sweeps/my_sweep_name-params_48000_20_20000_10_-6_0.5_0.5_cosine_1_10.wav"

The sweep parameters are encoded into the filename so an instance of an identical sweep, including the computed parameters and inverse filter, can be re-created automatically from the filename. Measurements can be deconvolved directly from file.

new_swept_sine = SweptSine.init_from_sweep_filename(
    "sweeps/my_sweep_name-params_params_48000_20_20000_10_-6_0.5_0.5_cosine_1_10.wav"
)

impulse_response = new_swept_sine.deconvolve_from_wav("measurement.wav")

Convolution

The SweptSine.convolve function can be used for auralisation, for example using impulse responses from the OpenAIR database.

auralised_signal = SweptSine.convolve(anechoic_signal, impulse_response)