Loudness: File-based or hardware processing?
How many broadcast and post engineers or operators are concerned about loudness correction in real time versus file-based? There is a common misconception that loudness correction whether it is done by in hardware or software, provides the same quality. This is a subject of frequent discussion with our customers.
For example, an engineer from a broadcast facility gave us some files and asked us to check them. When we looked at the files and corrected them, they seemed fine. So we asked him, why these files? And he said that these files had caused a problem when they were corrected for loudness.
When processing happens in a hardware-based environment, audio gets modified whether it needs to or not. The original sound design can’t be maintained.
In this particular case, the quiet bit became a lot louder to where even a bird tweeting outside was audibly more noticeable, and the camera person’s movements became apparent. The hardware processing created audible artifacts that rendered the audio useless.
When we measured the file, we found it was not a “problem file” at all. It came within loudness spec using our software processing and everything was fine.
The way Engine and Eff work is that they measure the program loudness (average value of audio content, and simply use global gain or attenuation to fix the file. If there’s a true peak error in the original file, or if one is created as a result of us applying gain, we create a gate around the peak and then apply local attenuation inside the gate. The advantage of this approach is that a minimum of processing is applied and the sound design is maintained.