Audio Quality on the Web

I just downloaded the new Miro player (formally Democracy Player). It's very nice, seems to organize videos reasonably well, is free, multi-platform, open source and includes features I believe all video players should provide.

But, when I got to checking it out, I was disappointed with the audio quality on many of the videos featured on Miro's Getting Started page. No fault goes to Miro as they simply aggregate videos from around the world. Many of the media programs contained audio of unacceptable technical quality. Audio levels were either too hot (see Diggnation below) or were recorded in ways that rendered them unacceptable for listening. It was if the producer merely viewed the video ignoring the audio before releasing.

As an example, listen to the audio of this Diggnation video. Here's a snippet of the waveform (click on image to enlarge):

Remarkably, while the Diggnation sample was shipped way too hot, the program material does not clip often because of this (yeah, hard to believe by looking at the waveform.) More disturbing, however, is that the program is littered with examples of how not to set the mic preamplifier - and we do find distorted audio resulting from this oversight. Maybe the engineer never heard of the AGC button. I do not advocate the use of AGC, but at least it will prevent overloading amplifier stages in the camera. If the audio was recorded separately, then they would undoubtedly have adequate equipment to monitor the audio gain at various stages. There would be no excuse.

Now, compare the Diggnation wavefore to the technically superior audio from an ABC News program on Miro:

The good news is that with a little knob twiddlin', watching the meters, and LISTENING to your program, audio quality can be reasonable even with the cheap cameras and audio gear that most use on vlogs these days.

How can you improve your audio for video? Well... it shouldn't be any different than audio for any other medium. I advocate reference levels to be the same as the DVD standard and the levels proposed in the AES Technical Group for Games: -18dBFS. That means the Diggnation audio is way too hot. Go back and compare the two and take note of the amplitude level ruler on the left side of each waveform.

For some tips and the definition of the -18dBFS proposal, see my post The State of Podcasts and Vlogs