EdwardHayashi-2023
commited on
Commit
•
d62f361
1
Parent(s):
749f4ae
Create AESDD.py
Browse files
AESDD.py
ADDED
@@ -0,0 +1,151 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Copyright 2022 The PolyAI and 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 |
+
# Standard Libraries
|
17 |
+
import datasets
|
18 |
+
import os
|
19 |
+
#------------------------------------------------------------------------------
|
20 |
+
"""Acted Emotional Speech Dynamic Database v1.0"""
|
21 |
+
|
22 |
+
_CITATION = """\
|
23 |
+
@article{vryzas2018speech,
|
24 |
+
title={Speech emotion recognition for performance interaction},
|
25 |
+
author={Vryzas, Nikolaos and Kotsakis, Rigas and Liatsou, Aikaterini and Dimoulas, Charalampos A and Kalliris, George},
|
26 |
+
journal={Journal of the Audio Engineering Society},
|
27 |
+
volume={66},
|
28 |
+
number={6},
|
29 |
+
pages={457--467},
|
30 |
+
year={2018},
|
31 |
+
publisher={Audio Engineering Society}
|
32 |
+
}
|
33 |
+
"""
|
34 |
+
|
35 |
+
_DESCRIPTION = """\
|
36 |
+
AESDD v1.0 was created on October 2017 in the Laboratory of Electronic Media,
|
37 |
+
School of Journalism and Mass Communications, Aristotle University of Thessaloniki,
|
38 |
+
for the needs of Speech Emotion Recognition research of the Multidisciplinary Media &
|
39 |
+
Mediated Communication Research Group (M3C, http://m3c.web.auth.gr/).
|
40 |
+
|
41 |
+
For the creation of v.1 of the database, 5 (3 female and 2 male) professional actors were recorded.
|
42 |
+
19 utterances of ambiguous out of context emotional content were chosen.
|
43 |
+
The actors acted these 19 utterances in every one of the 5 chosen emotions.
|
44 |
+
One extra improvised utterance was added for every actor and emotion.
|
45 |
+
The guidance of the actors and the choice of the final recordings were supervised by
|
46 |
+
a scientific expert in dramatology. For some of the utterances, more that one takes were qualified.
|
47 |
+
Consequently, around 500 utterances occured in the final database.
|
48 |
+
"""
|
49 |
+
|
50 |
+
_HOMEPAGE = "http://m3c.web.auth.gr/research/aesdd-speech-emotion-recognition/"
|
51 |
+
|
52 |
+
_LICENSE = "CC BY 4.0"
|
53 |
+
|
54 |
+
_DATA_URL = "https://drive.google.com/uc?export=download&id=1-pelMaCrfwoUCmwxUtlacRUBwbFnXlXA"
|
55 |
+
|
56 |
+
#------------------------------------------------------------------------------
|
57 |
+
# Define Dataset Configuration (e.g., subset of dataset, but it is not used here.)
|
58 |
+
class AESDDConfig(datasets.BuilderConfig):
|
59 |
+
#--------------------------------------------------------------------------
|
60 |
+
def __init__(self, name, description, homepage, data_url):
|
61 |
+
|
62 |
+
super(AESDDConfig, self).__init__(
|
63 |
+
name = self.name,
|
64 |
+
version = datasets.Version("1.0.0"),
|
65 |
+
description = self.description,
|
66 |
+
)
|
67 |
+
self.name = name
|
68 |
+
self.description = description
|
69 |
+
self.homepage = homepage
|
70 |
+
self.data_url = data_url
|
71 |
+
#------------------------------------------------------------------------------
|
72 |
+
# Define Dataset Class
|
73 |
+
class AESDD(datasets.GeneratorBasedBuilder):
|
74 |
+
#--------------------------------------------------------------------------
|
75 |
+
BUILDER_CONFIGS = [AESDDConfig(
|
76 |
+
name = "AESDD",
|
77 |
+
description = _DESCRIPTION,
|
78 |
+
homepage = _HOMEPAGE,
|
79 |
+
data_url = _DATA_URL
|
80 |
+
)]
|
81 |
+
#--------------------------------------------------------------------------
|
82 |
+
'''
|
83 |
+
Define the "column header" (feature) of a datum.
|
84 |
+
3 Features:
|
85 |
+
1) path_to_file
|
86 |
+
2) audio samples
|
87 |
+
3) emotion label
|
88 |
+
4) utterance: 1,2,...,20
|
89 |
+
5) speaker id
|
90 |
+
'''
|
91 |
+
def _info(self):
|
92 |
+
|
93 |
+
features = datasets.Features(
|
94 |
+
{
|
95 |
+
"path": datasets.Value("string"),
|
96 |
+
"audio": datasets.Audio(sampling_rate = 441000),
|
97 |
+
"label": datasets.ClassLabel(
|
98 |
+
names = [
|
99 |
+
"anger",
|
100 |
+
"disgust",
|
101 |
+
"fear",
|
102 |
+
"happiness",
|
103 |
+
"sadness",
|
104 |
+
]),
|
105 |
+
"utterance": datasets.Value("float"),
|
106 |
+
"speaker": datasets.Value("float")
|
107 |
+
}
|
108 |
+
)
|
109 |
+
|
110 |
+
# return dataset info and data feature info
|
111 |
+
return datasets.DatasetInfo(
|
112 |
+
description = _DESCRIPTION,
|
113 |
+
features = features,
|
114 |
+
homepage = _HOMEPAGE,
|
115 |
+
citation = _CITATION,
|
116 |
+
)
|
117 |
+
#--------------------------------------------------------------------------
|
118 |
+
def _split_generators(self, dl_manager):
|
119 |
+
|
120 |
+
dataset_path = dl_manager.download_and_extract(self.config.data_url)
|
121 |
+
|
122 |
+
return [
|
123 |
+
datasets.SplitGenerator(
|
124 |
+
# set the whole dataset as "training set". No worry, can split later!
|
125 |
+
name = datasets.Split.TRAIN,
|
126 |
+
# _generate_examples()'s parameters, thus name must match!
|
127 |
+
gen_kwargs = {
|
128 |
+
"dataset_path": dataset_path
|
129 |
+
},
|
130 |
+
)
|
131 |
+
]
|
132 |
+
#--------------------------------------------------------------------------
|
133 |
+
def _generate_examples(self, dataset_path):
|
134 |
+
'''
|
135 |
+
Get the audio file and set the corresponding labels
|
136 |
+
'''
|
137 |
+
key = 0
|
138 |
+
for dir_name in ["anger", "disgust", "fear", "happiness", "sadness"]:
|
139 |
+
dir_path = dataset_path + "/AESDD/" + dir_name
|
140 |
+
for file_name in os.listdir(dir_path):
|
141 |
+
if file_name.endswith(".wav"):
|
142 |
+
yield key, {
|
143 |
+
"path": dir_path + "/" + file_name,
|
144 |
+
# huggingface dataset's will use soundfile to read the audio file
|
145 |
+
"audio": dir_path + "/" + file_name,
|
146 |
+
"label": dir_name,
|
147 |
+
"utterance": float(file_name[1:3]),
|
148 |
+
"speaker": float(file_name[file_name.find("(")+1:file_name.find(")")]),
|
149 |
+
}
|
150 |
+
key += 1
|
151 |
+
#------------------------------------------------------------------------------
|