Abstract: Techniques for packet-accurate targeted content substitution by a stitcher are described herein. In some embodiments, the stitcher at a client device receives a live transport stream (TS) and a splice message indicating a start time and an end time of a splice period. The stitcher locates in the live TS a video packet with a video packet identifier (PID) and an audio packet with an audio PID crossing the start time. The stitcher injects to the live TS targeted content TS packets within the splice period, where the injection includes re-stamping PIDs and presentation timestamps of the targeted content TS packets to match the live TS packets, and re-stamping the live TS packets with a predefined PID starting from the video packet and the audio packet, and ceasing the re-stamping when a respective audio packet in the live TS packets crossing the end time.