5fa1a76
1
2
To identify unsupported tokens, extract all unique characters in the dataset using the SpeechT5Tokenizer which works with characters as tokens.
