Does anyone happen to know if ChatGPT's voice feature uses audio compression similar to Opus? Especially the "heavy 30 percent receiver-side packet loss" exampl

Does anyone happen to know if ChatGPT's voice feature uses audio compression similar to Opus? Especially the "heavy 30 percent receiver-side packet loss" example sounds a LOT like the experience I have sometimes.