Commit 55afb266 authored by Jan Kiene's avatar Jan Kiene
Browse files

add multiple formats in postprocessing 

- allows pre-conversion to MASA for using masaRenderer
parent 9d2a29fc

experiments/selection/P800-8/config/P800-8.yml

+4 −4
Original line number Diff line number Diff line
@@ -302,6 +302,6 @@ conditions_to_generate: 
### Post-processing

################################################

postprocessing:

    fmt: "BINAURAL"

    fmt: ["MASA2", "BINAURAL"]

    fs: 48000

    loudness: -26

ivas_processing_scripts/audiotools/convert/__init__.py

+13 −3
Original line number Diff line number Diff line
@@ -34,6 +34,7 @@ import logging 
from pathlib import Path, PurePath

from typing import Optional, Union

from numpy import empty

from shutil import copyfile



from ivas_processing_scripts.audiotools import audio, audioarray, metadata

from ivas_processing_scripts.audiotools.audiofile import write
@@ -100,7 +101,7 @@ def convert_file( 
        output.audio = empty((1, num_tcs))



        # fabricate metadata file name

        output.metadata_files = [Path(out_file).with_suffix(".met")]

        output.metadata_file = Path(out_file).with_suffix(".met")

    if isinstance(output, audio.ObjectBasedAudio):

        try:

            output.object_pos = input.object_pos
@@ -135,6 +136,11 @@ def convert_file( 
    write(out_file, output.audio, output.fs)

    if isinstance(output, audio.ObjectBasedAudio):

        write_ISM_metadata_in_file(output.object_pos, [out_file], automatic_naming=True)

    elif isinstance(output, audio.MetadataAssistedSpatialAudio) and in_fmt == out_fmt:

        # audio objects point to same MD file, create new one with default naming for output

        out_md_name = out_file.parent / (out_file.name + ".met")

        copyfile(output.metadata_file, out_md_name)

        output.metadata_file = out_md_name





def convert(
@@ -303,7 +309,7 @@ def format_conversion( 
    """Convert one audio format to another"""



    # validation

    if isinstance(output, audio.MetadataAssistedSpatialAudio) and not isinstance(input, audio.SceneBasedAudio):

    if isinstance(output, audio.MetadataAssistedSpatialAudio) and not (isinstance(input, audio.SceneBasedAudio) or isinstance(input, audio.MetadataAssistedSpatialAudio )):

        raise NotImplementedError("Can only convert to MASA from SBA")



    if isinstance(output, audio.ObjectBasedAudio) and input.name != output.name:
@@ -314,10 +320,14 @@ def format_conversion( 
    if logger:

        logger.debug(f"Format conversion: {input.name} -> {output.name}")



    if input.name == output.name or (

    if ( fmt := input.name ) == output.name or (

        input.name.startswith("BINAURAL") and output.name.startswith("BINAURAL")

    ):

        output.audio = input.audio

        if fmt.startswith("MASA"):

            output.metadata_file = input.metadata_file

        elif fmt.startswith("ISM"):

            output.metadata_files = list(output.metadata_files)

    else:

        if isinstance(input, audio.BinauralAudio):

            raise NotImplementedError(

ivas_processing_scripts/audiotools/convert/scenebased.py

+1 −1
Original line number Diff line number Diff line
@@ -192,7 +192,7 @@ def render_sba_to_masa( 
    # two dir only possible from HOA2

    num_dirs = 1

    num_tcs = masa_out.audio.shape[1]

    md_out_path = masa_out.metadata_files[0]

    md_out_path = masa_out.metadata_file

    

    masa = masaAnalyzer(sba_in, num_tcs, num_dirs, md_out_path)

    masa_out.audio = masa.audio

ivas_processing_scripts/processing/chains.py

+20 −1
Original line number Diff line number Diff line
@@ -409,13 +409,32 @@ def get_processing_chain( 
        loudness_postprocessing = post_cfg.get("loudness")

        loudness_fmt_postprocessing = post_cfg.get("loudness_fmt")



    post_fmt = post_cfg.get("fmt")

    if isinstance(post_fmt, list):

        pre_fmts = post_fmt[:-1]

        post_fmt = post_fmt[-1]



        # add Postprocessing with only format conversion for each format except the last

        fmts = [tmp_in_fmt] + pre_fmts

        for fmt_in, fmt_out in zip(fmts[:-1], fmts[1:]):

            chain["processes"].append(

                Postprocessing(

                    {

                        "in_fs": tmp_in_fs,

                        "in_fmt": fmt_in,

                        "out_fs": tmp_in_fs,

                        "out_fmt": fmt_out

                    }

                )

            )



    chain["processes"].append(

        Postprocessing(

            {

                "in_fs": tmp_in_fs,

                "in_fmt": tmp_in_fmt,

                "out_fs": post_cfg.get("fs"),

                "out_fmt": post_cfg.get("fmt"),

                "out_fmt": post_fmt,

                "out_cutoff": tmp_lp_cutoff,

                "out_loudness": loudness_postprocessing,

                "out_loudness_fmt": loudness_fmt_postprocessing,

ivas_processing_scripts/processing/processing.py

+15 −5
Original line number Diff line number Diff line
@@ -176,6 +176,8 @@ def concat_teardown(cfg: TestConfig, logger: logging.Logger): 
        raise ValueError("Splitting not possible without split marker")



    output_format = cfg.postprocessing["fmt"]

    if isinstance(output_format, list):

        output_format = output_format[-1]



    out_files = []

    out_meta = []
@@ -327,6 +329,10 @@ def reverse_process_2(cfg, logger): 
        logger.info("Remove preamble")

        remove_preamble(cfg)



    fmt = cfg.postprocessing["fmt"]

    if isinstance(fmt, list):

        fmt = fmt[-1]



    # reverse concatenation

    if cfg.pre2.concatenate_input:

        # write out the splits, optionally remove file
@@ -337,13 +343,13 @@ def reverse_process_2(cfg, logger): 
        for out_dir in cfg.out_dirs:

            list_audio_dir = list_audio(out_dir)

            out_paths_splits.append(list_audio_dir)

        if cfg.postprocessing["fmt"].startswith("ISM"):

        if fmt.startswith("ISM"):

            out_meta_splits = []

            for i, condition in enumerate(out_paths_splits):

                meta_condition = metadata_search(

                    cfg.out_dirs[i],

                    condition,

                    num_objects=int(cfg.postprocessing["fmt"][-1]),

                    num_objects=int(fmt[-1]),

                )

                out_meta_splits.append(meta_condition)

        else:
@@ -353,7 +359,7 @@ def reverse_process_2(cfg, logger): 
    if cfg.postprocessing.get("loudness", False):

        scale_files(

            out_paths_splits,

            cfg.postprocessing["fmt"],

            fmt,

            cfg.postprocessing["loudness"],

            cfg.postprocessing.get("loudness_fmt", None),

            cfg.postprocessing["fs"],
@@ -444,13 +450,17 @@ def process_item( 


def remove_preamble(cfg):

    # get number of channels from output format

    num_channels = audio.fromtype(cfg.postprocessing["fmt"]).num_channels

    fmt = cfg.postprocessing["fmt"]

    if isinstance(cfg.postprocessing["fmt"], list):

        fmt = fmt[-1]



    num_channels = audio.fromtype(fmt).num_channels

    for odir in cfg.out_dirs:

        for item in cfg.items_list:

            path_input = odir / item.name



            # remove preamble for ISM metadata

            if cfg.postprocessing["fmt"].startswith("ISM"):

            if fmt.startswith("ISM"):

                # search for metadata

                meta_item = metadata_search(

                    odir, [Path(item.name)], num_objects=num_channels