Fix for audio files without transcript

tv3_parla.py

@@ -104,6 +104,8 @@ class Tv3Parla(datasets.GeneratorBasedBuilder):
         for key, (path, file) in enumerate(audio_files):
             if path.endswith(".wav") and f"/{split}/" in path:
                 uid = path.split("/")[-1][:-4]
+                if uid not in transcripts:
+                    continue
                 text = transcripts.pop(uid)
                 audio = {"path": path, "bytes": file.read()}
                 yield key, {"path": path, "audio": audio, "text": text}