pyFDN.fdn_build_gallery

pyFDN.fdn_build_gallery(N=None, *, fs=48000.0, delays=None, delay_range=(400, 1200), delay_distribution='uniform', coprime=False, sort_delays=False, num_inputs=1, num_outputs=1, io_type='normalised', input_scale=1.0, output_scale=1.0, direct_gain=0.0, rt=2.0, rt_nyquist=None, rt_crossover=None, output_gain_db=None, output_gain_db_nyquist=None, output_crossover=None, rng=None, return_design=False)

Build a complete FDN and optionally return its non-inferable design.

Overloads:

• N (int | None), fs (float), delays (np.ndarray | None), delay_range (tuple[int, int]), delay_distribution (DelayDistribution), coprime (bool), sort_delays (bool), num_inputs (int), num_outputs (int), io_type (str), input_scale (float), output_scale (float), direct_gain (float | None), rt (float | None), rt_nyquist (float | None), rt_crossover (float | None), output_gain_db (ArrayLike | None), output_gain_db_nyquist (ArrayLike | None), output_crossover (float | None), rng (np.random.Generator | int | None), return_design (Literal[False]) → FDNBuild

• N (int | None), fs (float), delays (np.ndarray | None), delay_range (tuple[int, int]), delay_distribution (DelayDistribution), coprime (bool), sort_delays (bool), num_inputs (int), num_outputs (int), io_type (str), input_scale (float), output_scale (float), direct_gain (float | None), rt (float | None), rt_nyquist (float | None), rt_crossover (float | None), output_gain_db (ArrayLike | None), output_gain_db_nyquist (ArrayLike | None), output_crossover (float | None), rng (np.random.Generator | int | None), return_design (Literal[True]) → tuple[FDNBuild, FDNDesign]

The feedback matrix is random orthogonal. Sampled delays support the same distributions and coprimality option as sample_delay_lengths(). Explicit delays remain purely numerical and therefore produce no delay design record.

In-loop attenuation and optional output EQ are first-order shelves. Set return_design=True to return (build, design) for direct use in an pyFDN.FDNPreset; the default remains the plain FDNBuild.