xref: /aosp_15_r20/external/libopus/dnn/torch/osce/stndrd/evaluation/run_nomad.py (revision a58d3d2adb790c104798cd88c8a3aff4fa8b82cc)
1"""
2/* Copyright (c) 2023 Amazon
3   Written by Jan Buethe */
4/*
5   Redistribution and use in source and binary forms, with or without
6   modification, are permitted provided that the following conditions
7   are met:
8
9   - Redistributions of source code must retain the above copyright
10   notice, this list of conditions and the following disclaimer.
11
12   - Redistributions in binary form must reproduce the above copyright
13   notice, this list of conditions and the following disclaimer in the
14   documentation and/or other materials provided with the distribution.
15
16   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
17   ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
18   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
19   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
20   OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
21   EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
22   PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
23   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
24   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
25   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
26   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27*/
28"""
29
30import os
31import argparse
32import tempfile
33import shutil
34
35import pandas as pd
36from scipy.spatial.distance import cdist
37from scipy.io import wavfile
38import numpy as np
39
40from nomad_audio.nomad import Nomad
41
42
43parser = argparse.ArgumentParser()
44parser.add_argument('folder', type=str, help='folder with processed items')
45parser.add_argument('--full-reference', action='store_true', help='use NOMAD as full-reference metric')
46parser.add_argument('--device', type=str, default=None, help='device for Nomad')
47
48
49def get_bitrates(folder):
50    with open(os.path.join(folder, 'bitrates.txt')) as f:
51        x = f.read()
52
53    bitrates = [int(y) for y in x.rstrip('\n').split()]
54
55    return bitrates
56
57def get_itemlist(folder):
58    with open(os.path.join(folder, 'items.txt')) as f:
59        lines = f.readlines()
60
61    items = [x.split()[0] for x in lines]
62
63    return items
64
65
66def nomad_wrapper(ref_folder, deg_folder, full_reference=False, ref_embeddings=None, device=None):
67    model = Nomad(device=device)
68    if not full_reference:
69        results = model.predict(nmr=ref_folder, deg=deg_folder)[0].to_dict()['NOMAD']
70        return results, None
71    else:
72        if ref_embeddings is None:
73            print(f"Computing reference embeddings from {ref_folder}")
74            ref_data = pd.DataFrame(sorted(os.listdir(ref_folder)))
75            ref_data.columns = ['filename']
76            ref_data['filename'] = [os.path.join(ref_folder, x) for x in ref_data['filename']]
77            ref_embeddings = model.get_embeddings_csv(model.model, ref_data).set_index('filename')
78
79        print(f"Computing degraded embeddings from {deg_folder}")
80        deg_data = pd.DataFrame(sorted(os.listdir(deg_folder)))
81        deg_data.columns = ['filename']
82        deg_data['filename'] = [os.path.join(deg_folder, x) for x in deg_data['filename']]
83        deg_embeddings = model.get_embeddings_csv(model.model, deg_data).set_index('filename')
84
85        dist = np.diag(cdist(ref_embeddings, deg_embeddings)) # wasteful
86        test_files = [x.split('/')[-1].split('.')[0] for x in deg_embeddings.index]
87
88        results = dict(zip(test_files, dist))
89
90        return results, ref_embeddings
91
92
93
94
95def nomad_process_all(folder, full_reference=False, device=None):
96    bitrates = get_bitrates(folder)
97    items = get_itemlist(folder)
98    with tempfile.TemporaryDirectory() as dir:
99        cleandir  = os.path.join(dir, 'clean')
100        opusdir   = os.path.join(dir, 'opus')
101        lacedir   = os.path.join(dir, 'lace')
102        nolacedir = os.path.join(dir, 'nolace')
103
104        # prepare files
105        for d in [cleandir, opusdir, lacedir, nolacedir]: os.makedirs(d)
106        for br in bitrates:
107            for item in items:
108                for cond in ['clean', 'opus', 'lace', 'nolace']:
109                    shutil.copyfile(os.path.join(folder, cond, f"{item}_{br}_{cond}.wav"), os.path.join(dir, cond, f"{item}_{br}.wav"))
110
111        nomad_opus, ref_embeddings   = nomad_wrapper(cleandir, opusdir, full_reference=full_reference, ref_embeddings=None)
112        nomad_lace, ref_embeddings   = nomad_wrapper(cleandir, lacedir, full_reference=full_reference, ref_embeddings=ref_embeddings)
113        nomad_nolace, ref_embeddings = nomad_wrapper(cleandir, nolacedir, full_reference=full_reference, ref_embeddings=ref_embeddings)
114
115    results = dict()
116    for br in bitrates:
117        results[br] = np.zeros((len(items), 3))
118        for i, item in enumerate(items):
119            key = f"{item}_{br}"
120            results[br][i, 0] = nomad_opus[key]
121            results[br][i, 1] = nomad_lace[key]
122            results[br][i, 2] = nomad_nolace[key]
123
124    return results
125
126
127
128if __name__ == "__main__":
129    args = parser.parse_args()
130
131    items = get_itemlist(args.folder)
132    bitrates = get_bitrates(args.folder)
133
134    results = nomad_process_all(args.folder, full_reference=args.full_reference, device=args.device)
135
136    np.save(os.path.join(args.folder, f'results_nomad.npy'), results)
137
138    print("Done.")
139