holylovenia
commited on
Commit
•
1830e50
1
Parent(s):
2f8675a
Upload asr_ibsc.py with huggingface_hub
Browse files- asr_ibsc.py +190 -0
asr_ibsc.py
ADDED
@@ -0,0 +1,190 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
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 |
+
from pathlib import Path
|
17 |
+
from typing import Dict, List, Tuple
|
18 |
+
|
19 |
+
import datasets
|
20 |
+
import fsspec
|
21 |
+
import pandas as pd
|
22 |
+
from fsspec.callbacks import TqdmCallback
|
23 |
+
|
24 |
+
from seacrowd.utils.configs import SEACrowdConfig
|
25 |
+
from seacrowd.utils.constants import (SCHEMA_TO_FEATURES, TASK_TO_SCHEMA,
|
26 |
+
Licenses, Tasks)
|
27 |
+
|
28 |
+
_CITATION = """\
|
29 |
+
@inproceedings{Juan14,
|
30 |
+
Title = {Semi-supervised G2P bootstrapping and its application to ASR for a very under-resourced language: Iban},
|
31 |
+
Author = {Sarah Samson Juan and Laurent Besacier and Solange Rossato},
|
32 |
+
Booktitle = {Proceedings of Workshop for Spoken Language Technology for Under-resourced (SLTU)},
|
33 |
+
Year = {2014}}
|
34 |
+
Month = {May},
|
35 |
+
|
36 |
+
@inproceedings{Juan2015,
|
37 |
+
Title = {Using Resources from a closely-Related language to develop ASR for a very under-resourced Language: A case study for Iban},
|
38 |
+
Author = {Sarah Samson Juan and Laurent Besacier and Benjamin Lecouteux and Mohamed Dyab},
|
39 |
+
Booktitle = {Proceedings of INTERSPEECH},
|
40 |
+
Year = {2015},
|
41 |
+
Month = {September}}
|
42 |
+
Address = {Dresden, Germany},
|
43 |
+
"""
|
44 |
+
|
45 |
+
_DATASETNAME = "asr_ibsc"
|
46 |
+
|
47 |
+
_DESCRIPTION = """\
|
48 |
+
This package contains Iban language text and speech suitable for Automatic
|
49 |
+
Speech Recognition (ASR) experiments. In addition, transcribed speech, 2M tokens
|
50 |
+
corpus crawled from an online newspaper site is provided. News data was provided
|
51 |
+
by a local radio station in Sarawak, Malaysia.
|
52 |
+
"""
|
53 |
+
|
54 |
+
_HOMEPAGE = "https://github.com/sarahjuan/iban"
|
55 |
+
|
56 |
+
_LANGUAGES = ["iba"]
|
57 |
+
|
58 |
+
_LICENSE = Licenses.CC_BY_SA_3_0.value
|
59 |
+
|
60 |
+
_LOCAL = False
|
61 |
+
|
62 |
+
_URL = "https://github.com/sarahjuan/iban/tree/master/data"
|
63 |
+
|
64 |
+
_SUPPORTED_TASKS = [Tasks.SPEECH_RECOGNITION]
|
65 |
+
_SEACROWD_SCHEMA = f"seacrowd_{TASK_TO_SCHEMA[_SUPPORTED_TASKS[0]].lower()}" # sptext
|
66 |
+
|
67 |
+
_SOURCE_VERSION = "1.0.0"
|
68 |
+
|
69 |
+
_SEACROWD_VERSION = "2024.06.20"
|
70 |
+
|
71 |
+
|
72 |
+
class ASRIbanDataset(datasets.GeneratorBasedBuilder):
|
73 |
+
"""Iban language text and speech suitable for ASR experiments"""
|
74 |
+
|
75 |
+
SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
|
76 |
+
SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)
|
77 |
+
|
78 |
+
BUILDER_CONFIGS = [
|
79 |
+
SEACrowdConfig(
|
80 |
+
name=f"{_DATASETNAME}_source",
|
81 |
+
version=SOURCE_VERSION,
|
82 |
+
description=f"{_DATASETNAME} source schema",
|
83 |
+
schema="source",
|
84 |
+
subset_id=_DATASETNAME,
|
85 |
+
),
|
86 |
+
SEACrowdConfig(
|
87 |
+
name=f"{_DATASETNAME}_{_SEACROWD_SCHEMA}",
|
88 |
+
version=SEACROWD_VERSION,
|
89 |
+
description=f"{_DATASETNAME} SEACrowd schema",
|
90 |
+
schema=_SEACROWD_SCHEMA,
|
91 |
+
subset_id=_DATASETNAME,
|
92 |
+
),
|
93 |
+
]
|
94 |
+
|
95 |
+
DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"
|
96 |
+
|
97 |
+
def _info(self) -> datasets.DatasetInfo:
|
98 |
+
if self.config.schema == "source":
|
99 |
+
features = datasets.Features(
|
100 |
+
{
|
101 |
+
"audio": datasets.Audio(sampling_rate=16_000),
|
102 |
+
"transcription": datasets.Value("string"),
|
103 |
+
"speaker_id": datasets.Value("string"),
|
104 |
+
}
|
105 |
+
)
|
106 |
+
elif self.config.schema == _SEACROWD_SCHEMA:
|
107 |
+
features = SCHEMA_TO_FEATURES[TASK_TO_SCHEMA[_SUPPORTED_TASKS[0]]] # speech_text_features
|
108 |
+
|
109 |
+
return datasets.DatasetInfo(
|
110 |
+
description=_DESCRIPTION,
|
111 |
+
features=features,
|
112 |
+
homepage=_HOMEPAGE,
|
113 |
+
license=_LICENSE,
|
114 |
+
citation=_CITATION,
|
115 |
+
)
|
116 |
+
|
117 |
+
def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
|
118 |
+
"""Returns SplitGenerators."""
|
119 |
+
# prepare data directory
|
120 |
+
data_dir = Path.cwd() / "data" / "asr_ibsc"
|
121 |
+
data_dir.mkdir(parents=True, exist_ok=True)
|
122 |
+
|
123 |
+
# download data
|
124 |
+
# if rate limiting is an issue, pass github username and token
|
125 |
+
username = None
|
126 |
+
token = None
|
127 |
+
fs = fsspec.filesystem("github", org="sarahjuan", repo="iban", ref="master", username=username, token=token)
|
128 |
+
fs.clear_instance_cache()
|
129 |
+
|
130 |
+
# download annotation
|
131 |
+
print("Downloading annotation...")
|
132 |
+
fs.get(fs.ls("data/train/"), (data_dir / "train").as_posix(), recursive=True)
|
133 |
+
fs.get(fs.ls("data/test/"), (data_dir / "test").as_posix(), recursive=True)
|
134 |
+
|
135 |
+
# download audio files
|
136 |
+
print("Downloading audio files (~1GB). It may take several minutes...")
|
137 |
+
for idx, folder in enumerate(fs.ls("data/wav/")):
|
138 |
+
folder_name = folder.split("/")[-1]
|
139 |
+
pbar = TqdmCallback(tqdm_kwargs={"desc": f"-> {folder_name} [{idx+1:2d}/{len(fs.ls('data/wav/'))}]", "unit": "file"})
|
140 |
+
fs.get(fs.ls(f"data/wav/{folder_name}/"), (data_dir / "wav" / folder_name).as_posix(), recursive=True, callback=pbar)
|
141 |
+
|
142 |
+
return [
|
143 |
+
datasets.SplitGenerator(
|
144 |
+
name=datasets.Split.TRAIN,
|
145 |
+
gen_kwargs={
|
146 |
+
"data_dir": data_dir,
|
147 |
+
"split": "train",
|
148 |
+
},
|
149 |
+
),
|
150 |
+
datasets.SplitGenerator(
|
151 |
+
name=datasets.Split.TEST,
|
152 |
+
gen_kwargs={
|
153 |
+
"data_dir": data_dir,
|
154 |
+
"split": "test",
|
155 |
+
},
|
156 |
+
),
|
157 |
+
]
|
158 |
+
|
159 |
+
def _generate_examples(self, data_dir: Path, split: str) -> Tuple[int, Dict]:
|
160 |
+
"""Yields examples as (key, example) tuples."""
|
161 |
+
|
162 |
+
text_file = data_dir / split / f"{split}_text"
|
163 |
+
utt2spk_file = data_dir / split / f"{split}_utt2spk"
|
164 |
+
wav_scp_file = data_dir / split / f"{split}_wav.scp"
|
165 |
+
|
166 |
+
# load the data
|
167 |
+
text_df = pd.read_csv(text_file, sep=" ", header=None, names=["utt_id", "text"])
|
168 |
+
utt2spk_df = pd.read_csv(utt2spk_file, sep="\t", header=None, names=["utt_id", "speaker"])
|
169 |
+
wav_df = pd.read_csv(wav_scp_file, sep="\t", header=None, names=["utt_id", "wav_path"])
|
170 |
+
merged_df = pd.merge(text_df, utt2spk_df, on="utt_id")
|
171 |
+
merged_df = pd.merge(merged_df, wav_df, on="utt_id")
|
172 |
+
|
173 |
+
for _, row in merged_df.iterrows():
|
174 |
+
wav_file = data_dir / "wav" / row["speaker"] / row["wav_path"].split("/")[-1]
|
175 |
+
|
176 |
+
if self.config.schema == "source":
|
177 |
+
yield row["utt_id"], {
|
178 |
+
"audio": str(wav_file.as_posix()),
|
179 |
+
"transcription": row["text"],
|
180 |
+
"speaker_id": row["speaker"],
|
181 |
+
}
|
182 |
+
elif self.config.schema == _SEACROWD_SCHEMA:
|
183 |
+
yield row["utt_id"], {
|
184 |
+
"id": row["utt_id"],
|
185 |
+
"path": str(wav_file),
|
186 |
+
"audio": str(wav_file.as_posix()),
|
187 |
+
"text": row["text"],
|
188 |
+
"speaker_id": row["speaker"],
|
189 |
+
"metadata": None,
|
190 |
+
}
|