get_tokens.py

from transformers import Wav2Vec2ProcessorWithLM
import torchaudio

import torch
from datasets import load_dataset
from transformers import AutoModelForCTC, AutoProcessor
import torchaudio.functional as F

model_id = "."
model = AutoModelForCTC.from_pretrained(model_id)
processor = AutoProcessor.from_pretrained(model_id)

vocab_dict = processor.tokenizer.get_vocab()
print(vocab_dict)
sorted_vocab_dict = {k.lower(): v for k, v in sorted(vocab_dict.items(), key=lambda item: item[1])}