Update NURC-SP_ENTOA_TTS.py

NURC-SP_ENTOA_TTS.py

@@ -2,6 +2,7 @@ import csv
 import datasets
 from datasets import BuilderConfig, GeneratorBasedBuilder, DatasetInfo, SplitGenerator, Split
 import logging
+from pathlib import Path
 
 # Set up logging
 logging.basicConfig(level=logging.INFO)
@@ -103,13 +104,17 @@ class NurcSPDataset(GeneratorBasedBuilder):
         examples = {}
         example_count = 0
         
-        # Read CSV file
+        # Read CSV file and log some sample paths
+        logger.info("Reading CSV file...")
         with open(prompts_path, "r") as f:
             csv_reader = csv.DictReader(f)
-            for row in csv_reader:
-                examples[row['file_path']] = {
+            for i, row in enumerate(csv_reader):
+                file_path = row['file_path']
+                # Normalize path separators
+                file_path = Path(file_path).as_posix()
+                examples[file_path] = {
                     "audio_name": row['audio_name'],
-                    "file_path": row['file_path'],
+                    "file_path": file_path,
                     "text": row['text'],
                     "start_time": row['start_time'],
                     "end_time": row['end_time'],
@@ -125,15 +130,25 @@ class NurcSPDataset(GeneratorBasedBuilder):
                     "speaker_id": row['speaker_id'],
                 }
                 example_count += 1
+                # Log first few paths for debugging
+                if i < 5:
+                    logger.info(f"Sample CSV path {i}: {file_path}")
         
         logger.info(f"Found {example_count} examples in CSV")
         
         inside_clips_dir = False
         id_ = 0
         matched_files = 0
+        archive_paths_seen = set()
+        
+        # Log the path_to_clips we're looking for
+        logger.info(f"Looking for files in directory: {path_to_clips}")
         
         for path, f in audio_files:
-            logger.debug(f"Processing archive file: {path}")
+            # Normalize path separators
+            path = Path(path).as_posix()
+            archive_paths_seen.add(path)
+            
             if path.startswith(path_to_clips):
                 inside_clips_dir = True
                 logger.debug(f"Found file in clips directory: {path}")
@@ -142,9 +157,20 @@ class NurcSPDataset(GeneratorBasedBuilder):
                     matched_files += 1
                     yield id_, {**examples[path], "audio": audio}
                     id_ += 1
+                else:
+                    # Log unmatched paths for first few files
+                    if len(archive_paths_seen) < 5:
+                        logger.warning(f"Archive path not found in CSV: {path}")
             elif inside_clips_dir:
                 break
         
+        # Log some sample archive paths for comparison
+        logger.info("Sample paths from archive:")
+        for path in list(archive_paths_seen)[:5]:
+            logger.info(f"Archive path: {path}")
+        
+        logger.info(f"Total archive paths found: {len(archive_paths_seen)}")
         logger.info(f"Matched {matched_files} audio files with CSV entries")
         if matched_files == 0:
-            logger.warning("No audio files were matched with CSV entries!")
+            logger.warning("No audio files were matched with CSV entries!")
+            logger.warning("This might be due to path mismatches. Check if the paths in your CSV match the structure in your .tar.gz file.")