Post
by Kai Stromler » Wed Oct 29, 2003 3:46 pm
My system, allowing complete range of scores 1-10. Note that this may not be applicable to others without as much audio experience.
10. I am sitting in the prod booth as the pre-master is being played back by the band, or in the front row five feet from a PA cabinet if the song is a live take. The song has obviously been recovered or enhanced beyond the limitations of the CD-audio format, which as audiophiles know are pretty hefty. Any editing is seamless and the result sounds and flows better than the original arrangement. Stereo channels are utilized to good effect. As would appear, I don't give out 10s often.
9. The audio sounds like a CD. I can hear two distinct stereo channels, even if it's largely the same sound in both. This is pretty much the default rating.
8. There is something wrong with the audio, but it's pretty minor. Perhaps it's been mixed weird to push up the mids as opposed to the highs and lows, or the original recording quality was poor. Very well done recoveries from old LPs will land around 8, because there's nearly always some noise associated with the analog format. Normal MP3s used as initial source usually top out around this level.
7. There is something wrong with the audio that shouldn't be, like clipped wave tops or a missing channel. This score is also associated with very clean movie-trailer audio, which will inevitably start 'crumpled' as a result of being from a distribution encode.
6. The original audio source is clearly sub-commercial; there is either unintended distortion or obvious audio artifacting which indicates that it is from a poorly encoded MP3. This level can also be earned by poor normalization: if it's inaudible or ear-bleeding at normal volumes and neither is useful to the video, it gets a 6.
5. The distortion patterns are characteristic of certain proprietary compression schemes. It's possible to make WMA and RealAudio sound good, but there was no effort put toward such here. Normal trailer sound also runs around 5 due to inevitable damage in the extraction and reencoding processes.
4. Multiple things are wrong here, such as bad normalization resulting in clipped waves. Many LPs and cassettes, due to age and media damage, will earn 4s without audio repair.
3. The audio sounds like the violent parts of Ulver's _Nattens Madrigal_, even though it wasn't the artist's original intent, nor the video's concept, to make the audio difficult to listen to. There was a piss-poor recording job done, and nothing done by the editor to clean up the soundprint. As an example, consider a song recorded by microphone from a playing radio, then converted to digital format without modification.
2. The audio seems to have been deliberately damaged. The volume is unpredictable, the signal cuts in and out, there is heavy static and distortion, and generally, the audio has been jumped on with football cleats. In this category are many public radio addresses released onto the web.
1. The audio track is either absent or completely full of sonic garbage, and there is evidence that the creator intended there to be music or other intelligible audio. Very few videos will get 1s, because sound this bad is almost always done so on purpose, and the audio will be reevaluated in the light of its effect on the overall concept.
I know that I've given at least one 10 and at least one 3, though no 2s or 1s yet, because, as described, it's almost as difficult to do those as it is to do enough mods to get a 10.
--K
