Datasets:

ArXiv:
License:
polinaeterna HF staff commited on
Commit
00846d9
·
1 Parent(s): a44f1ba
Files changed (1) hide show
  1. evi.py +23 -30
evi.py CHANGED
@@ -12,6 +12,12 @@
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 csv
17
  import json
@@ -23,8 +29,6 @@ import datasets
23
  logger = datasets.logging.get_logger(__name__)
24
 
25
 
26
- """ EVI Dataset"""
27
-
28
  _CITATION = """\
29
  @inproceedings{Spithourakis2022evi,
30
  author = {Georgios P. Spithourakis and Ivan Vuli\'{c} and Micha\l{} Lis and I\~{n}igo Casanueva and Pawe\l{} Budzianowski},
@@ -42,9 +46,15 @@ _ALL_CONFIGS = sorted([
42
 
43
  _LANGS = sorted(["en", "fr", "pl"])
44
 
45
- _DESCRIPTION = "EVI is a dataset for enrolment, identification, and verification" # noqa
 
 
 
 
46
 
47
- _HOMEPAGE_URL = "https://arxiv.org/abs/2204.13496"
 
 
48
 
49
  _BASE_URL = "https://huggingface.co/datasets/PolyAI/evi/resolve/main/data"
50
 
@@ -54,9 +64,7 @@ _TEXT_URL = {
54
 
55
  _AUDIO_DATA_URL = "https://poly-public-data.s3.eu-west-2.amazonaws.com/evi-paper/audios.zip" # noqa
56
 
57
- _ARCHIVE_PATH = "zip://::https://poly-public-data.s3.eu-west-2.amazonaws.com/evi-paper/audios.zip"
58
-
59
- _VERSION = datasets.Version("0.0.10", "")
60
 
61
 
62
  class EviConfig(datasets.BuilderConfig):
@@ -69,19 +77,10 @@ class EviConfig(datasets.BuilderConfig):
69
  self.languages = _LANGS if name == "all" else [name.split("-")[0]] # all langs if config == "all"
70
 
71
 
72
- def _build_config(name):
73
- return EviConfig(
74
- name=name,
75
- version=_VERSION,
76
- description=_DESCRIPTION,
77
- # homepage=_HOMEPAGE_URL,
78
- )
79
-
80
-
81
  class Evi(datasets.GeneratorBasedBuilder):
82
 
83
- DEFAULT_WRITER_BATCH_SIZE = 256
84
- BUILDER_CONFIGS = [_build_config(name) for name in _ALL_CONFIGS + ["all"]]
85
 
86
  def _info(self):
87
  features = datasets.Features(
@@ -99,18 +98,16 @@ class Evi(datasets.GeneratorBasedBuilder):
99
  )
100
 
101
  return datasets.DatasetInfo(
102
- version=self.config.version,
103
- description=self.config.description,
104
- license="CC-BY-4.0",
105
  citation=_CITATION,
106
  features=features,
 
107
  )
108
 
109
  def _split_generators(self, dl_manager):
110
-
111
  langs = self.config.languages
112
- # "./audios/{lang}/{dialogue_id}/3.wav"
113
-
114
  lang2text_urls = {
115
  lang: _TEXT_URL[lang] for lang in langs
116
  }
@@ -121,10 +118,8 @@ class Evi(datasets.GeneratorBasedBuilder):
121
  datasets.SplitGenerator(
122
  name=datasets.Split.TEST,
123
  gen_kwargs={
124
- # "audio_data": dl_manager.iter_files(audio_data_path),
125
  "audio_data_path": audio_data_path,
126
  "text_paths": lang2text_paths,
127
- # "streaming": dl_manager.is_streaming
128
  },
129
  )
130
  ]
@@ -137,9 +132,7 @@ class Evi(datasets.GeneratorBasedBuilder):
137
  reader = csv.DictReader(
138
  fin, delimiter="\t", skipinitialspace=True
139
  )
140
- for i, dictrow in enumerate(reader):
141
- if i == 199:
142
- break
143
  dialogue_id = dictrow["dialogue_id"]
144
  turn_id = dictrow["turn_num"]
145
  file_path = os.path.join(
@@ -148,7 +141,7 @@ class Evi(datasets.GeneratorBasedBuilder):
148
  dialogue_id,
149
  f'{turn_id}.wav'
150
  )
151
- full_path = Path(audio_data_path) / file_path
152
 
153
  yield file_path, {
154
  "language": lang,
 
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
+ EVI is a challenging spoken multilingual dataset with 5,506 dialogues in English, Polish, and French
17
+ that can be used for benchmarking and developing knowledge-based enrolment, identification, and identification
18
+ for spoken dialogue systems.
19
+ """
20
+
21
 
22
  import csv
23
  import json
 
29
  logger = datasets.logging.get_logger(__name__)
30
 
31
 
 
 
32
  _CITATION = """\
33
  @inproceedings{Spithourakis2022evi,
34
  author = {Georgios P. Spithourakis and Ivan Vuli\'{c} and Micha\l{} Lis and I\~{n}igo Casanueva and Pawe\l{} Budzianowski},
 
46
 
47
  _LANGS = sorted(["en", "fr", "pl"])
48
 
49
+ _DESCRIPTION = """
50
+ EVI is a challenging spoken multilingual dataset with 5,506 dialogues in English, Polish, and French
51
+ that can be used for benchmarking and developing knowledge-based enrolment, identification, and identification
52
+ for spoken dialogue systems.
53
+ """ # noqa
54
 
55
+ _LICENSE = "CC-BY-4.0"
56
+
57
+ _HOMEPAGE = "https://github.com/PolyAI-LDN/evi-paper"
58
 
59
  _BASE_URL = "https://huggingface.co/datasets/PolyAI/evi/resolve/main/data"
60
 
 
64
 
65
  _AUDIO_DATA_URL = "https://poly-public-data.s3.eu-west-2.amazonaws.com/evi-paper/audios.zip" # noqa
66
 
67
+ _VERSION = datasets.Version("0.0.1", "")
 
 
68
 
69
 
70
  class EviConfig(datasets.BuilderConfig):
 
77
  self.languages = _LANGS if name == "all" else [name.split("-")[0]] # all langs if config == "all"
78
 
79
 
 
 
 
 
 
 
 
 
 
80
  class Evi(datasets.GeneratorBasedBuilder):
81
 
82
+ DEFAULT_WRITER_BATCH_SIZE = 512
83
+ BUILDER_CONFIGS = [EviConfig(name) for name in _ALL_CONFIGS + ["all"]]
84
 
85
  def _info(self):
86
  features = datasets.Features(
 
98
  )
99
 
100
  return datasets.DatasetInfo(
101
+ version=_VERSION,
102
+ description=_DESCRIPTION,
103
+ license=_LICENSE,
104
  citation=_CITATION,
105
  features=features,
106
+ homepage=_HOMEPAGE
107
  )
108
 
109
  def _split_generators(self, dl_manager):
 
110
  langs = self.config.languages
 
 
111
  lang2text_urls = {
112
  lang: _TEXT_URL[lang] for lang in langs
113
  }
 
118
  datasets.SplitGenerator(
119
  name=datasets.Split.TEST,
120
  gen_kwargs={
 
121
  "audio_data_path": audio_data_path,
122
  "text_paths": lang2text_paths,
 
123
  },
124
  )
125
  ]
 
132
  reader = csv.DictReader(
133
  fin, delimiter="\t", skipinitialspace=True
134
  )
135
+ for dictrow in reader:
 
 
136
  dialogue_id = dictrow["dialogue_id"]
137
  turn_id = dictrow["turn_num"]
138
  file_path = os.path.join(
 
141
  dialogue_id,
142
  f'{turn_id}.wav'
143
  )
144
+ full_path = os.path.join(audio_data_path, file_path)
145
 
146
  yield file_path, {
147
  "language": lang,