Audio Crackle AAC Audio (mp4) Video Exports

Bug
2

A few thoughts:

  1. Would it be possible to export from Audition as WAV instead of MP3? Aside from the lossy vs lossless benefits, WAV also helps because MP3 is not a seek-friendly format at the sample level. If Shotcut exports with Parallel Processing turned on (and sometimes even when off), this seek inaccuracy can cause sync problems.

  2. I assume the Audition mix targets 1 dB headroom because you know the side effects of quantization used by lossy formats like MP3. The lossy audio approximates the original source in a way that compresses better, but the approximation will sometimes be a little louder or softer than the original by virtue of being an approximation. When the approximation is louder, any original sound mixed to 0 dB goes into the clipping range, hence the need for 1 dB headroom. However, the workflow we see so far is an MP3 source being recompressed to an AAC target. That is double lossy compression and approximation, which means the MP3 headroom needs to be at least 1.5 dB. Or, the video project needs an output (stereo bus) limiter at 1 to 1.5 dB to protect the AAC approximation from clipping.

  3. AAC is far from perfect. Would AC-3 or MP3 be an option, or Opus in a Matroska container? Those codecs tend to be more faithful to the original sound. Here’s a discussion with distortion graphs that you may find interesting:

1 Like