Update training data description to 'publicly available recordings with aligned transcripts'

Remove processor_config.json (incompatible with transformers 4.44 — WhisperProcessor reconstructs from other files)

Add special_tokens_map.json from openai/whisper-small (standard Whisper tokenizer file)

Add added_tokens.json from openai/whisper-small (standard Whisper tokenizer file)

Add normalizer.json from openai/whisper-small (standard Whisper tokenizer file)

Add merges.txt from openai/whisper-small (standard Whisper tokenizer file)

Add vocab.json from openai/whisper-small (standard Whisper tokenizer file)

Add preprocessor_config.json from openai/whisper-small (WhisperFeatureExtractor config, identical across small fine-tunes)

Update model card with matched-eval kirtan metrics

Add model card

surt_small_v3 final (step 12000)

initial commit