In order to encode sound for streaming playback, we have to first decode it into an uncompressed format (we use WAV) and then re-encode as MP3. The issue is that there is a hard limit in the WAV specification, and there's no way for us or anybody else to get around it. When we decode a very large MP3 to WAV, and it goes over the limit, the audio usually gets truncated. 

 

Right now, it would be best to split the sound into two parts and upload each separately.