With suppress_non_speech_tokens set to true I'm still getting non speech tokens

[itsthisjustin]

Transcriptions are including things such as "Music Playing" and "Bell Rings". Am I missing something silly here or does this param just not work? Using the Swift package fyi

[KTibow]

only these are suppressed

static const std::vector<std::string> non_speech_tokens = {
    "\"", "#", "(", ")", "*", "+", "/", ":", ";", "<", "=", ">", "@", "[", "\\", "]", "^",
    "_", "`", "{", "|", "}", "~", "「", "」", "『", "』", "<<", ">>", "<<<", ">>>", "--",
    "---", "-(", "-[", "('", "(\"", "((", "))", "(((", ")))", "[[", "]]", "{{", "}}", "♪♪",
    "♪♪♪","♩", "♪", "♫", "♬", "♭", "♮", "♯"
};