Is this only in the app preview or also in the exported result played in a media player? Preview is not optimal because it is trying to do many things in real-time and introduces processing latencies.
Also, AAC is not optimal for editing because it does not always accurately cut to the correct sample. Uncompressed (WAV/PCM) or an audio codec with no delay such as AC-3 works better. However, it is possible the problem is some combination of audio channels and frame rate problem especially if you are changing the frame rate when you export, which you should avoid.