Datasets:
Tags:
speech-modeling
License:
Fixes sampling rate reporting for 16K_mp3
Browse files
NPSC.py
CHANGED
@@ -85,6 +85,7 @@ class Npsc(datasets.GeneratorBasedBuilder):
|
|
85 |
]
|
86 |
|
87 |
def _info(self):
|
|
|
88 |
return datasets.DatasetInfo(
|
89 |
description=_DESCRIPTION,
|
90 |
features=datasets.Features(
|
@@ -101,7 +102,7 @@ class Npsc(datasets.GeneratorBasedBuilder):
|
|
101 |
"normsentence_text": datasets.Value("string"),
|
102 |
"transsentence_text": datasets.Value("string"),
|
103 |
"translated": datasets.Value("int32"),
|
104 |
-
"audio": datasets.features.Audio(sampling_rate=
|
105 |
|
106 |
}
|
107 |
),
|
|
|
85 |
]
|
86 |
|
87 |
def _info(self):
|
88 |
+
sampling_rate = 16_000 if self.config.name == "16K_mp3" else 48_000
|
89 |
return datasets.DatasetInfo(
|
90 |
description=_DESCRIPTION,
|
91 |
features=datasets.Features(
|
|
|
102 |
"normsentence_text": datasets.Value("string"),
|
103 |
"transsentence_text": datasets.Value("string"),
|
104 |
"translated": datasets.Value("int32"),
|
105 |
+
"audio": datasets.features.Audio(sampling_rate=sampling_rate),
|
106 |
|
107 |
}
|
108 |
),
|