Update superb_demo.py
Browse files- superb_demo.py +19 -6
superb_demo.py
CHANGED
@@ -135,6 +135,7 @@ class Superb(datasets.GeneratorBasedBuilder):
|
|
135 |
features=datasets.Features(
|
136 |
{
|
137 |
"file": datasets.Value("string"),
|
|
|
138 |
"text": datasets.Value("string"),
|
139 |
"speaker_id": datasets.Value("int64"),
|
140 |
"chapter_id": datasets.Value("int64"),
|
@@ -159,6 +160,7 @@ class Superb(datasets.GeneratorBasedBuilder):
|
|
159 |
features=datasets.Features(
|
160 |
{
|
161 |
"file": datasets.Value("string"),
|
|
|
162 |
"label": datasets.ClassLabel(
|
163 |
names=[
|
164 |
"yes",
|
@@ -192,6 +194,7 @@ class Superb(datasets.GeneratorBasedBuilder):
|
|
192 |
features=datasets.Features(
|
193 |
{
|
194 |
"file": datasets.Value("string"),
|
|
|
195 |
"speaker_id": datasets.Value("string"),
|
196 |
"text": datasets.Value("string"),
|
197 |
"action": datasets.ClassLabel(
|
@@ -234,6 +237,7 @@ class Superb(datasets.GeneratorBasedBuilder):
|
|
234 |
features=datasets.Features(
|
235 |
{
|
236 |
"file": datasets.Value("string"),
|
|
|
237 |
"label": datasets.ClassLabel(names=[f"id{i + 10001}" for i in range(1251)]),
|
238 |
}
|
239 |
),
|
@@ -253,6 +257,7 @@ class Superb(datasets.GeneratorBasedBuilder):
|
|
253 |
features=datasets.Features(
|
254 |
{
|
255 |
"file": datasets.Value("string"),
|
|
|
256 |
"label": datasets.ClassLabel(names=['neu', 'hap', 'ang', 'sad']),
|
257 |
}
|
258 |
),
|
@@ -327,11 +332,13 @@ class Superb(datasets.GeneratorBasedBuilder):
|
|
327 |
id_, transcript = line.split(" ", 1)
|
328 |
audio_file = f"{id_}.flac"
|
329 |
speaker_id, chapter_id = [int(el) for el in id_.split("-")[:2]]
|
|
|
330 |
yield key, {
|
331 |
"id": id_,
|
332 |
"speaker_id": speaker_id,
|
333 |
"chapter_id": chapter_id,
|
334 |
-
"file":
|
|
|
335 |
"text": transcript,
|
336 |
}
|
337 |
key += 1
|
@@ -347,7 +354,7 @@ class Superb(datasets.GeneratorBasedBuilder):
|
|
347 |
label = "_silence_"
|
348 |
else:
|
349 |
label = "_unknown_"
|
350 |
-
yield key, {"file": audio_file, "label": label}
|
351 |
elif self.config.name == "ic":
|
352 |
root_path = os.path.join(archive_path, "fluent_speech_commands_dataset/")
|
353 |
csv_path = os.path.join(root_path, f"data/{split}_data.csv")
|
@@ -356,8 +363,10 @@ class Superb(datasets.GeneratorBasedBuilder):
|
|
356 |
next(csv_reader)
|
357 |
for row in csv_reader:
|
358 |
key, file_path, speaker_id, text, action, object_, location = row
|
|
|
359 |
yield key, {
|
360 |
-
"file":
|
|
|
361 |
"speaker_id": speaker_id,
|
362 |
"text": text,
|
363 |
"action": action,
|
@@ -373,8 +382,10 @@ class Superb(datasets.GeneratorBasedBuilder):
|
|
373 |
if int(split_id) != split:
|
374 |
continue
|
375 |
speaker_id = file_path.split("/")[0]
|
|
|
376 |
yield key, {
|
377 |
-
"file":
|
|
|
378 |
"label": speaker_id,
|
379 |
}
|
380 |
elif self.config.name == "er":
|
@@ -393,9 +404,11 @@ class Superb(datasets.GeneratorBasedBuilder):
|
|
393 |
continue
|
394 |
wav_subdir = filename.rsplit("_", 1)[0]
|
395 |
filename = f"{filename}.wav"
|
|
|
396 |
yield key, {
|
397 |
-
"file":
|
398 |
-
"
|
|
|
399 |
}
|
400 |
key += 1
|
401 |
|
|
|
135 |
features=datasets.Features(
|
136 |
{
|
137 |
"file": datasets.Value("string"),
|
138 |
+
"audio": datasets.features.Audio(sampling_rate=16_000),
|
139 |
"text": datasets.Value("string"),
|
140 |
"speaker_id": datasets.Value("int64"),
|
141 |
"chapter_id": datasets.Value("int64"),
|
|
|
160 |
features=datasets.Features(
|
161 |
{
|
162 |
"file": datasets.Value("string"),
|
163 |
+
"audio": datasets.features.Audio(sampling_rate=16_000),
|
164 |
"label": datasets.ClassLabel(
|
165 |
names=[
|
166 |
"yes",
|
|
|
194 |
features=datasets.Features(
|
195 |
{
|
196 |
"file": datasets.Value("string"),
|
197 |
+
"audio": datasets.features.Audio(sampling_rate=16_000),
|
198 |
"speaker_id": datasets.Value("string"),
|
199 |
"text": datasets.Value("string"),
|
200 |
"action": datasets.ClassLabel(
|
|
|
237 |
features=datasets.Features(
|
238 |
{
|
239 |
"file": datasets.Value("string"),
|
240 |
+
"audio": datasets.features.Audio(sampling_rate=16_000),
|
241 |
"label": datasets.ClassLabel(names=[f"id{i + 10001}" for i in range(1251)]),
|
242 |
}
|
243 |
),
|
|
|
257 |
features=datasets.Features(
|
258 |
{
|
259 |
"file": datasets.Value("string"),
|
260 |
+
"audio": datasets.features.Audio(sampling_rate=16_000),
|
261 |
"label": datasets.ClassLabel(names=['neu', 'hap', 'ang', 'sad']),
|
262 |
}
|
263 |
),
|
|
|
332 |
id_, transcript = line.split(" ", 1)
|
333 |
audio_file = f"{id_}.flac"
|
334 |
speaker_id, chapter_id = [int(el) for el in id_.split("-")[:2]]
|
335 |
+
audio_path = os.path.join(transcript_dir_path, audio_file)
|
336 |
yield key, {
|
337 |
"id": id_,
|
338 |
"speaker_id": speaker_id,
|
339 |
"chapter_id": chapter_id,
|
340 |
+
"file": audio_path,
|
341 |
+
"audio": audio_path,
|
342 |
"text": transcript,
|
343 |
}
|
344 |
key += 1
|
|
|
354 |
label = "_silence_"
|
355 |
else:
|
356 |
label = "_unknown_"
|
357 |
+
yield key, {"file": audio_file, "audio": audio_file, "label": label}
|
358 |
elif self.config.name == "ic":
|
359 |
root_path = os.path.join(archive_path, "fluent_speech_commands_dataset/")
|
360 |
csv_path = os.path.join(root_path, f"data/{split}_data.csv")
|
|
|
363 |
next(csv_reader)
|
364 |
for row in csv_reader:
|
365 |
key, file_path, speaker_id, text, action, object_, location = row
|
366 |
+
audio_path = os.path.join(root_path, file_path)
|
367 |
yield key, {
|
368 |
+
"file": audio_path,
|
369 |
+
"audio": audio_path,
|
370 |
"speaker_id": speaker_id,
|
371 |
"text": text,
|
372 |
"action": action,
|
|
|
382 |
if int(split_id) != split:
|
383 |
continue
|
384 |
speaker_id = file_path.split("/")[0]
|
385 |
+
audio_path = os.path.join(wav_path, file_path)
|
386 |
yield key, {
|
387 |
+
"file": audio_path,
|
388 |
+
"audio": audio_path,
|
389 |
"label": speaker_id,
|
390 |
}
|
391 |
elif self.config.name == "er":
|
|
|
404 |
continue
|
405 |
wav_subdir = filename.rsplit("_", 1)[0]
|
406 |
filename = f"{filename}.wav"
|
407 |
+
audio_path = os.path.join(wav_path, wav_subdir, filename)
|
408 |
yield key, {
|
409 |
+
"file": audio_path,
|
410 |
+
"audio": audio_path,
|
411 |
+
"label": emo.replace("exc", "hap"),
|
412 |
}
|
413 |
key += 1
|
414 |
|