init
Browse files- push_s2t_translation.py +5 -26
push_s2t_translation.py
CHANGED
@@ -59,12 +59,14 @@ for i in tqdm(list(range(line_no_start, line_no_end))):
|
|
59 |
continue
|
60 |
i = loader(files[i])
|
61 |
i[f"{direction_text}.text"] = line2text[str(i["line_no"])]
|
62 |
-
audio_file = i
|
63 |
start, end = i[f"{direction_speech}.duration_start"], i[f"{direction_speech}.duration_end"]
|
64 |
if os.path.exists(audio_file):
|
65 |
try:
|
66 |
wav = audio_loader.decode_example({"path": audio_file, "bytes": None})
|
67 |
if start < end < len(wav["array"]):
|
|
|
|
|
68 |
features.append(i)
|
69 |
else:
|
70 |
delete_audio(audio_file)
|
@@ -74,34 +76,11 @@ for i in tqdm(list(range(line_no_start, line_no_end))):
|
|
74 |
|
75 |
|
76 |
print(f"features (filtered): {len(features)}")
|
77 |
-
data_dict = {f"{direction_speech}.audio": [i.pop(f"{direction_speech}.path") for i in features]}
|
78 |
keys = features[0].keys()
|
79 |
-
data_dict
|
80 |
audio_dataset = Dataset.from_dict(data_dict)
|
81 |
audio_dataset = audio_dataset.cast_column(f"{direction_speech}.audio", Audio())
|
82 |
-
|
83 |
-
|
84 |
-
# trim the audio according to the duration
|
85 |
-
def clip_audio(batch):
|
86 |
-
start = batch[f"{direction_speech}.duration_start"]
|
87 |
-
end = batch[f"{direction_speech}.duration_end"]
|
88 |
-
audio = batch[f"{direction_speech}.audio"]
|
89 |
-
batch[f"{direction_speech}.audio"] = [
|
90 |
-
{"array": a["array"][s:e], "sampling_rate": a["sampling_rate"]}
|
91 |
-
for a, s, e in zip(audio, start, end)
|
92 |
-
]
|
93 |
-
return batch
|
94 |
-
|
95 |
-
|
96 |
-
audio_dataset_valid = audio_dataset_valid.map(
|
97 |
-
function=clip_audio,
|
98 |
-
batched=True,
|
99 |
-
batch_size=128,
|
100 |
-
num_proc=1,
|
101 |
-
desc="clipping audio based on the duration:"
|
102 |
-
)
|
103 |
-
|
104 |
-
dataset_to_push = DatasetDict({"train": audio_dataset_valid})
|
105 |
repo_name = f"{hf_org}/{hf_dataset}"
|
106 |
while True:
|
107 |
try:
|
|
|
59 |
continue
|
60 |
i = loader(files[i])
|
61 |
i[f"{direction_text}.text"] = line2text[str(i["line_no"])]
|
62 |
+
audio_file = i.pop(f"{direction_speech}.path")
|
63 |
start, end = i[f"{direction_speech}.duration_start"], i[f"{direction_speech}.duration_end"]
|
64 |
if os.path.exists(audio_file):
|
65 |
try:
|
66 |
wav = audio_loader.decode_example({"path": audio_file, "bytes": None})
|
67 |
if start < end < len(wav["array"]):
|
68 |
+
wav["array"] = wav["array"][start:end]
|
69 |
+
i[f"{direction_speech}.audio"] = wav
|
70 |
features.append(i)
|
71 |
else:
|
72 |
delete_audio(audio_file)
|
|
|
76 |
|
77 |
|
78 |
print(f"features (filtered): {len(features)}")
|
|
|
79 |
keys = features[0].keys()
|
80 |
+
data_dict = {k: [i[k] for i in features] for k in keys}
|
81 |
audio_dataset = Dataset.from_dict(data_dict)
|
82 |
audio_dataset = audio_dataset.cast_column(f"{direction_speech}.audio", Audio())
|
83 |
+
dataset_to_push = DatasetDict({"train": audio_dataset})
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
84 |
repo_name = f"{hf_org}/{hf_dataset}"
|
85 |
while True:
|
86 |
try:
|