Datasets:
Should path be a feature?
Browse files
rixvox.py
CHANGED
@@ -84,8 +84,8 @@ class Rixvox(datasets.GeneratorBasedBuilder):
|
|
84 |
"end": datasets.Value("float64"),
|
85 |
"duration": datasets.Value("float64"),
|
86 |
"bleu_score": datasets.Value("float64"),
|
87 |
-
|
88 |
-
|
89 |
"speaker_total_hours": datasets.Value("float64"),
|
90 |
# These are the features of your dataset like images, labels ...
|
91 |
}
|
@@ -176,7 +176,7 @@ class Rixvox(datasets.GeneratorBasedBuilder):
|
|
176 |
else:
|
177 |
df_meta = pd.read_parquet(meta_paths[self.config.name])
|
178 |
|
179 |
-
df_meta = df_meta.set_index("filename")
|
180 |
|
181 |
for i, audio_archive in enumerate(archive_iters):
|
182 |
for filename, file in audio_archive:
|
|
|
84 |
"end": datasets.Value("float64"),
|
85 |
"duration": datasets.Value("float64"),
|
86 |
"bleu_score": datasets.Value("float64"),
|
87 |
+
"filename": datasets.Value("string"),
|
88 |
+
"path": datasets.Value("string"),
|
89 |
"speaker_total_hours": datasets.Value("float64"),
|
90 |
# These are the features of your dataset like images, labels ...
|
91 |
}
|
|
|
176 |
else:
|
177 |
df_meta = pd.read_parquet(meta_paths[self.config.name])
|
178 |
|
179 |
+
df_meta = df_meta.set_index("filename", drop=False)
|
180 |
|
181 |
for i, audio_archive in enumerate(archive_iters):
|
182 |
for filename, file in audio_archive:
|