Comment by samcrawford

Do you have any intuition for whether the echhoprint-codegen algorithm would be suitable for saying whether two voice recordings match? One would be a little lossy, the other pretty much perfect.