There is provided mechanisms for decoding a media stream. A method is performed by a packet receiver. The method comprises receiving packets. The packets have been transmitted by a packet transmitter. The packets define the media stream and each packet comprises encoded media. The method comprises decoding the encoded media into at least a first decoded version and a second decoded version. For the first decoded version only a subset of the packets decoded for the second decoded version are decoded. The subset includes only those of the packets that are received within a first threshold time delay from having been transmitted by the packet transmitter.