incercarea #20

horoscop_neti.py

@@ -0,0 +1,76 @@
+import csv
+import os
+import json
+
+import datasets
+from datasets.utils.py_utils import size_str
+from tqdm import tqdm
+
+# _BASE_URL = "https://huggingface.co/datasets/mozilla-foundation/common_voice_11_0/resolve/main/"
+
+class HoroscopNeti(datasets.GeneratorBasedBuilder):
+    
+    def _info(self):
+        return datasets.DatasetInfo(
+            description="Descrierea datasetului tău.",
+            features=datasets.Features({
+                "path": datasets.Value("string"),
+                "audio": datasets.features.Audio(sampling_rate=16000),
+                "transcript": datasets.Value("string"),  
+            }),
+            supervised_keys=("audio", "transcript"),
+            homepage="URL către pagina de Hugging Face a datasetului",
+            citation="Referința de citare a datasetului",
+        )
+
+    def _split_generators(self):
+
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={
+                    "audio_dir": "audio/train",
+                    "transcript_dir": "transcript/train",
+                },
+            ),
+            datasets.SplitGenerator(
+                name=datasets.Split.TEST,
+                gen_kwargs={
+                    "audio_dir": "audio/test",
+                    "transcript_dir": "transcript/test",
+                },
+            ),
+            datasets.SplitGenerator(
+                name=datasets.Split.VALIDATION,
+                gen_kwargs={
+                    "audio_dir": "audio/validation",
+                    "transcript_dir": "transcript/validation",
+                },
+            ),
+        ]
+
+    def _generate_examples(self, audio_dir, transcript_dir):
+        for root, _, files in os.walk(transcript_dir):
+            for file in files:
+                if not file.endswith(".txt"):
+                    continue
+                    
+                transcript_path = os.path.join(transcript_dir, file)
+                audio_path = os.path.join(audio_dir, file.replace(".txt", ".wav"))
+                
+                if not os.path.exists(audio_path):
+                    continue
+                
+                with open(transcript_path, "r", encoding="utf-8") as f:
+                    transcript = f.read().strip()
+                
+                with open(audio_path, "rb") as af:
+                    audio_bytes = af.read()
+                
+                example_id = os.path.splitext(file)[0]
+                
+                yield example_id, {
+                    "path": audio_path,
+                    "audio": {"path": audio_path, "bytes": audio_bytes},
+                    "transcript": transcript,
+                }