updated _generate_examples
Browse files- common_voice_bn.py +116 -39
common_voice_bn.py
CHANGED
@@ -226,55 +226,132 @@ class CommonVoice(datasets.GeneratorBasedBuilder):
|
|
226 |
# ),
|
227 |
]
|
228 |
|
229 |
-
|
230 |
-
|
231 |
-
def _generate_examples(
|
232 |
self,
|
233 |
local_extracted_archive,
|
234 |
archive_iterator,
|
235 |
metadata_filepath,
|
236 |
path_to_clips,
|
|
|
237 |
):
|
238 |
"""Yields examples."""
|
239 |
data_fields = list(self._info().features.keys())
|
240 |
metadata = {}
|
241 |
metadata_found = False
|
242 |
-
|
243 |
-
|
244 |
-
|
245 |
-
|
246 |
-
|
247 |
-
|
248 |
-
|
249 |
-
|
250 |
-
row["path"]
|
251 |
-
|
252 |
-
|
253 |
-
|
254 |
-
|
255 |
-
|
256 |
-
|
257 |
-
|
258 |
-
|
259 |
-
|
260 |
-
|
261 |
-
|
262 |
-
|
263 |
-
|
264 |
-
|
265 |
-
|
266 |
-
|
267 |
-
|
268 |
-
|
269 |
-
os.path.join(local_extracted_archive, path)
|
270 |
-
|
271 |
-
|
272 |
-
|
273 |
-
|
274 |
-
|
275 |
-
|
276 |
-
|
277 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
278 |
|
279 |
|
280 |
|
|
|
226 |
# ),
|
227 |
]
|
228 |
|
229 |
+
def _generate_examples(
|
|
|
|
|
230 |
self,
|
231 |
local_extracted_archive,
|
232 |
archive_iterator,
|
233 |
metadata_filepath,
|
234 |
path_to_clips,
|
235 |
+
mode
|
236 |
):
|
237 |
"""Yields examples."""
|
238 |
data_fields = list(self._info().features.keys())
|
239 |
metadata = {}
|
240 |
metadata_found = False
|
241 |
+
if mode in ["dev","test"]:
|
242 |
+
for path, f in archive_iterator:
|
243 |
+
if path == metadata_filepath:
|
244 |
+
metadata_found = True
|
245 |
+
lines = (line.decode("utf-8") for line in f)
|
246 |
+
reader = csv.DictReader(lines, delimiter="\t", quoting=csv.QUOTE_NONE)
|
247 |
+
for row in reader:
|
248 |
+
# set absolute path for mp3 audio file
|
249 |
+
if not row["path"].endswith(".mp3"):
|
250 |
+
row["path"] += ".mp3"
|
251 |
+
row["path"] = os.path.join(path_to_clips, row["path"])
|
252 |
+
# accent -> accents in CV 8.0
|
253 |
+
if "accents" in row:
|
254 |
+
row["accent"] = row["accents"]
|
255 |
+
del row["accents"]
|
256 |
+
# if data is incomplete, fill with empty values
|
257 |
+
for field in data_fields:
|
258 |
+
if field not in row:
|
259 |
+
row[field] = ""
|
260 |
+
metadata[row["path"]] = row
|
261 |
+
elif path.startswith(path_to_clips):
|
262 |
+
assert metadata_found, "Found audio clips before the metadata TSV file."
|
263 |
+
if not metadata:
|
264 |
+
break
|
265 |
+
if path in metadata:
|
266 |
+
result = metadata[path]
|
267 |
+
# set the audio feature and the path to the extracted file
|
268 |
+
path = os.path.join(local_extracted_archive, path) if local_extracted_archive else path
|
269 |
+
result["audio"] = {"path": path, "bytes": f.read()}
|
270 |
+
# set path to None if the audio file doesn't exist locally (i.e. in streaming mode)
|
271 |
+
result["path"] = path if local_extracted_archive else None
|
272 |
+
|
273 |
+
yield path, result
|
274 |
+
else:
|
275 |
+
metadata_found = True
|
276 |
+
with open(metadata_filepath, "rb") as file_obj:
|
277 |
+
lines = (line.decode("utf-8") for line in file_obj)
|
278 |
+
reader = csv.DictReader(lines, delimiter="\t", quoting=csv.QUOTE_NONE)
|
279 |
+
for row in reader:
|
280 |
+
# set absolute path for mp3 audio file
|
281 |
+
if not row["path"].endswith(".mp3"):
|
282 |
+
row["path"] += ".mp3"
|
283 |
+
row["path"] = os.path.join(path_to_clips, row["path"])
|
284 |
+
# accent -> accents in CV 8.0
|
285 |
+
if "accents" in row:
|
286 |
+
row["accent"] = row["accents"]
|
287 |
+
del row["accents"]
|
288 |
+
# if data is incomplete, fill with empty values
|
289 |
+
for field in data_fields:
|
290 |
+
if field not in row:
|
291 |
+
row[field] = ""
|
292 |
+
metadata[row["path"]] = row
|
293 |
+
for path, f in archive_iterator:
|
294 |
+
if path.startswith(path_to_clips):
|
295 |
+
assert metadata_found, "Found audio clips before the metadata TSV file."
|
296 |
+
if not metadata:
|
297 |
+
break
|
298 |
+
if path in metadata:
|
299 |
+
result = metadata[path]
|
300 |
+
# set the audio feature and the path to the extracted file
|
301 |
+
path = os.path.join(local_extracted_archive, path) if local_extracted_archive else path
|
302 |
+
result["audio"] = {"path": path, "bytes": f.read()}
|
303 |
+
# set path to None if the audio file doesn't exist locally (i.e. in streaming mode)
|
304 |
+
result["path"] = path if local_extracted_archive else None
|
305 |
+
|
306 |
+
yield path, result
|
307 |
+
|
308 |
+
# def _generate_examples(
|
309 |
+
# self,
|
310 |
+
# local_extracted_archive,
|
311 |
+
# archive_iterator,
|
312 |
+
# metadata_filepath,
|
313 |
+
# path_to_clips,
|
314 |
+
# ):
|
315 |
+
# """Yields examples."""
|
316 |
+
# data_fields = list(self._info().features.keys())
|
317 |
+
# metadata = {}
|
318 |
+
# metadata_found = False
|
319 |
+
# for path, f in archive_iterator:
|
320 |
+
# if path == metadata_filepath:
|
321 |
+
# metadata_found = True
|
322 |
+
# lines = (line.decode("utf-8") for line in f)
|
323 |
+
# reader = csv.DictReader(lines, delimiter="\t", quoting=csv.QUOTE_NONE)
|
324 |
+
# for row in reader:
|
325 |
+
# # set absolute path for mp3 audio file
|
326 |
+
# if not row["path"].endswith(".mp3"):
|
327 |
+
# row["path"] += ".mp3"
|
328 |
+
# row["path"] = os.path.join(path_to_clips, row["path"])
|
329 |
+
# # accent -> accents in CV 8.0
|
330 |
+
# if "accents" in row:
|
331 |
+
# row["accent"] = row["accents"]
|
332 |
+
# del row["accents"]
|
333 |
+
# # if data is incomplete, fill with empty values
|
334 |
+
# for field in data_fields:
|
335 |
+
# if field not in row:
|
336 |
+
# row[field] = ""
|
337 |
+
# metadata[row["path"]] = row
|
338 |
+
# elif path.startswith(path_to_clips):
|
339 |
+
# assert metadata_found, "Found audio clips before the metadata TSV file."
|
340 |
+
# if not metadata:
|
341 |
+
# break
|
342 |
+
# if path in metadata:
|
343 |
+
# result = metadata[path]
|
344 |
+
# # set the audio feature and the path to the extracted file
|
345 |
+
# path = (
|
346 |
+
# os.path.join(local_extracted_archive, path)
|
347 |
+
# if local_extracted_archive
|
348 |
+
# else path
|
349 |
+
# )
|
350 |
+
# result["audio"] = {"path": path, "bytes": f.read()}
|
351 |
+
# # set path to None if the audio file doesn't exist locally (i.e. in streaming mode)
|
352 |
+
# result["path"] = path if local_extracted_archive else None
|
353 |
+
|
354 |
+
# yield path, result
|
355 |
|
356 |
|
357 |
|