Commit 5796a0e6 authored by Anika Treffehn's avatar Anika Treffehn
Browse files

preprocessing background noise

parent f5038c0c

ivas_processing_scripts/processing/chains.py

+1 −0
Original line number Diff line number Diff line
@@ -139,6 +139,7 @@ def get_preprocessing_2(cfg: TestConfig) -> dict: 
            "seed_delay": background_cfg.get("seed_delay", 0),

            "master_seed": cfg.master_seed,

            "output_fmt": cfg.postprocessing["fmt"],

            "background_object": None,

        }

    else:

        background = None

ivas_processing_scripts/processing/preprocessing_2.py

+1 −6
Original line number Diff line number Diff line
@@ -105,12 +105,7 @@ class Preprocessing2(Processing): 
        range_delay = (1, 2400000)



        # load background noise

        noise_object = audio.fromfile(

            self.in_fmt,

            self.background_noise["background_noise_path"],

            fs=self.in_fs,

            in_meta=in_meta,

        )

        noise_object = self.background_noise["background_object"]



        # if noise is too short raise error

        if len(noise_object.audio) < len(audio_object.audio):

ivas_processing_scripts/processing/processing.py

+17 −7
Original line number Diff line number Diff line
@@ -457,15 +457,25 @@ def remove_preamble(cfg): 




def preprocess_background_noise(cfg):

    # TODO

    # set values

    #out_format = # if preprocessing formt else input format

    #out_fs = # if preprocessing fs else input fs



    # TODO: add checks and errors for sampling rate and number channels compared to input signals

    # create audio objects

    input_audio = audio.fromfile(cfg.input["fmt"], cfg.pre.background_noise["background_noise_path"], fs=cfg.input.get("fs", None))

    output_audio = audio.fromtype(cfg.out_fmt)

    input_audio = audio.fromfile(cfg.input["fmt"], cfg.pre2.background_noise["background_noise_path"], fs=cfg.input.get("fs", None))

    output_audio = audio.fromtype(cfg.pre.out_fmt)

    output_audio.fs = input_audio.fs

    if cfg.pre.out_fs is None:

        out_fs = input_audio.fs

    else:

        out_fs = cfg.pre.out_fs



    # only consider format conversion, resampling and high-pass filtering

    convert(input=input_audio, output=output_audio, in_fs=cfg.input.get("fmt", None), out_fs=0, in_hp50=0)

    convert(input=input_audio, output=output_audio, in_fs=input_audio.fs, out_fs=out_fs, in_hp50=cfg.pre.in_hp50)



    # save result in cfg

    # cfg.preprocessing_2["background_noise"].update({"background_object": output_audio})

    cfg.proc_chains[0]["processes"][0].background_noise["background_object"] = output_audio



    return