Developers can now build fast speech-to-speech experiences into their applications