This paper investigates the benefits of streaming autonomous audio objects over error-prone channels instead of encoded audio frames. Due to the nature of autonomous audio objects such a scheme is error resilient and has a fine-grain scalable bitrate, but also has the additional benefit of being able to disguise packet loss in the reconstructed signal. This paper proposes object-packing algorithms which will be shown to be able to disguise the presence of long bursts of packet loss, removing the need for complex error-concealment schemes at the decoder.