myleslinder commited on
Commit
904cb61
·
1 Parent(s): 6ce3ece
Files changed (3) hide show
  1. .gitignore +4 -0
  2. crema-d.py +164 -0
  3. data/crema_d.tar.gz +3 -0
.gitignore ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ .DS_Store
2
+ Pipfile
3
+ Pipfile.lock
4
+ example.py
crema-d.py ADDED
@@ -0,0 +1,164 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ # Copyright 2022 The HuggingFace Datasets Authors and the current dataset script contributor.
3
+ #
4
+ # Licensed under the Apache License, Version 2.0 (the "License");
5
+ # you may not use this file except in compliance with the License.
6
+ # You may obtain a copy of the License at
7
+ #
8
+ # http://www.apache.org/licenses/LICENSE-2.0
9
+ #
10
+ # Unless required by applicable law or agreed to in writing, software
11
+ # distributed under the License is distributed on an "AS IS" BASIS,
12
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13
+ # See the License for the specific language governing permissions and
14
+ # limitations under the License.
15
+
16
+ import os
17
+ import datasets # type: ignore
18
+
19
+ logger = datasets.logging.get_logger(__name__)
20
+
21
+
22
+ """CREMA-D (Crowd-sourced Emotional Multimodal Actors Dataset)"""
23
+
24
+ _CITATION = """\
25
+ @article{cao2014crema,
26
+ title={CREMA-D: Crowd-sourced Emotional Multimodal Actors Dataset},
27
+ author={Cao, H. and Cooper, D. G. and Keutmann, M. K. and Gur, R. C. and Nenkova, A. and Verma, R.},
28
+ journal={IEEE transactions on affective computing},
29
+ volume={5},
30
+ number={4},
31
+ pages={377--390},
32
+ year={2014},
33
+ doi={10.1109/TAFFC.2014.2336244},
34
+ url={https://doi.org/10.1109/TAFFC.2014.2336244}
35
+ }
36
+ """
37
+
38
+ _DESCRIPTION = """\
39
+ CREMA-D is a data set of 7,442 original clips from 91 actors.
40
+ These clips were from 48 male and 43 female actors between the ages of 20 and 74
41
+ coming from a variety of races and ethnicities (African America, Asian, Caucasian, Hispanic, and Unspecified).
42
+ Actors spoke from a selection of 12 sentences.
43
+ The sentences were presented using one of six different emotions (Anger, Disgust, Fear, Happy, Neutral and Sad)
44
+ and four different emotion levels (Low, Medium, High and Unspecified).
45
+ Participants rated the emotion and emotion levels based on the combined audiovisual presentation,
46
+ the video alone, and the audio alone. Due to the large number of ratings needed, this effort was crowd-sourced
47
+ and a total of 2443 participants each rated 90 unique clips, 30 audio, 30 visual, and 30 audio-visual.
48
+ 95% of the clips have more than 7 rating.
49
+ """
50
+
51
+ _HOMEPAGE = "https://github.com/CheyneyComputerScience/CREMA-D"
52
+ _LICENSE = "ODbL"
53
+
54
+ _ROOT_DIR = "crema_d"
55
+ _DATA_URL = f"data/{_ROOT_DIR}.tar.gz"
56
+
57
+
58
+ _SENTENCE_MAP = {
59
+ "IEO": "It's eleven o'clock",
60
+ "TIE": "That is exactly what happened",
61
+ "IOM": "I'm on my way to the meeting",
62
+ "IWW": "I wonder what this is about",
63
+ "TAI": "The airplane is almost full",
64
+ "MTI": "Maybe tomorrow it will be cold",
65
+ "IWL": "I would like a new alarm clock",
66
+ "ITH": "I think I have a doctor's appointment",
67
+ "DFA": "Don't forget a jacket",
68
+ "ITS": "I think I've seen this before",
69
+ "TSI": "The surface is slick",
70
+ "WSI": "We'll stop in a couple of minutes",
71
+ }
72
+
73
+
74
+ _EMOTION_MAP = {
75
+ "ANG": "anger",
76
+ "DIS": "disgust",
77
+ "FEA": "fear",
78
+ "HAP": "happy",
79
+ "NEU": "neutral",
80
+ "SAD": "sad",
81
+ }
82
+
83
+ _INTENSITY_MAP = {
84
+ "LO": "Low",
85
+ "MD": "Medium",
86
+ "HI": "High",
87
+ "XX": "Unspecified",
88
+ ## on stray file
89
+ "X": "Unspecified",
90
+ }
91
+
92
+ _CLASS_NAMES = list(_EMOTION_MAP.values())
93
+
94
+
95
+ class CremaDDataset(datasets.GeneratorBasedBuilder):
96
+ """The Crema-D dataset"""
97
+
98
+ VERSION = datasets.Version("1.0.0")
99
+
100
+ def _info(self):
101
+ sampling_rate = 16_000
102
+ features = datasets.Features(
103
+ {
104
+ # "path": datasets.Value("string"),
105
+ "audio": datasets.Audio(sampling_rate=sampling_rate),
106
+ "actor_id": datasets.Value("string"),
107
+ "sentence": datasets.Value("string"),
108
+ # "emotion": datasets.Value("string"),
109
+ "emotion_intensity": datasets.Value("string"),
110
+ "label": datasets.ClassLabel(names=_CLASS_NAMES),
111
+ }
112
+ )
113
+
114
+ return datasets.DatasetInfo(
115
+ description=_DESCRIPTION,
116
+ features=features,
117
+ homepage=_HOMEPAGE,
118
+ citation=_CITATION,
119
+ license=_LICENSE,
120
+ # task_templates=[datasets.TaskTemplate("audio-classification")],
121
+ )
122
+
123
+ def _split_generators(self, dl_manager):
124
+
125
+ archive = dl_manager.download(_DATA_URL)
126
+ local_extracted_archive = (
127
+ dl_manager.extract(archive) if not dl_manager.is_streaming else None
128
+ )
129
+
130
+ return [
131
+ datasets.SplitGenerator(
132
+ name=datasets.Split.TRAIN,
133
+ gen_kwargs={
134
+ # "archive_path": _ROOT_DIR,
135
+ "local_extracted_archive": local_extracted_archive,
136
+ "audio_files": dl_manager.iter_archive(archive),
137
+ },
138
+ )
139
+ ]
140
+
141
+ def _generate_examples(self, local_extracted_archive, audio_files):
142
+ "4digitActorId_sentenceId_emotionId_emotionLevel"
143
+
144
+ id_ = 0
145
+ for path, f in audio_files:
146
+ path = os.path.join(
147
+ local_extracted_archive, path
148
+ ) # if local_extracted_archive else path
149
+ filename = os.path.basename(path)
150
+ with open(path, "rb") as f:
151
+ audio_bytes = f.read()
152
+ actor_id, sentence_id, emotion_id, emotion_level = filename.split(".")[
153
+ 0
154
+ ].split("_")
155
+ base = {
156
+ "path": path,
157
+ "actor_id": actor_id,
158
+ "sentence": _SENTENCE_MAP[sentence_id],
159
+ "label": _EMOTION_MAP[emotion_id],
160
+ "emotion_intensity": _INTENSITY_MAP[emotion_level],
161
+ }
162
+ audio = {"path": path, "bytes": audio_bytes}
163
+ yield id_, {**base, "audio": audio}
164
+ id_ += 1
data/crema_d.tar.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87b9dcce1fdcad68f8769201e047b505ff31cb38c782a6757aa2fb59c31e5590
3
+ size 470756748