I don’t have an audio snippet, but I can transcribe what the voice says on different runs. It usually pronounces random letters individually, but sometimes pronounces syllables with letters missing:

“L-W-R-D”, “L-L-W-R-L”, “hell-erl”, “H-L-er-D””, “eor-D”, “H-L-L-erl”, “L-L-W-L-R-D”, “H-L-W-R-L”, “hell-W-R”