Is there a theoretical maximum of 'over' that can be detected with oversampling? I'm asking this, because I stumbled on an MP3 file that shows outrageous overs in digicheck. It just tickles my curiosity.
Digicheck shows a peak of +4,5 dBfs in the left channel and +1,9 dBfs right.
Is a number of +4,5 dBfs plausible at all?
For the curious: The file in question is a rip of "The Gossip"s Album "Music for Men", the track's name is "Heavy Cross". It is encoded at 256kbit/s and 44.1kHz SR. The decoding is through Winamp with 24bit enabled and output using ASIO plugin from otachan. This ASIO plugin circumvents all level control within Winamp, so I am pretty sure that Winamp is not to blame here. I am using an analog attenuator to control my volume.
That is sadly clearly possible I'm afraid.
A lot to read here:
http://www.tcelectronic.com/loudness/li … -glossary/
And the cheat here:
One of the biggest problems with intersample overs and true peak overs giving the +0dBFS problem when MP3/AAC etc. is made from a file where the engineer doesn't have control.
In fact, today's lossy compresion codecs is making the loudness problem even worse.
To make it even worse, the DA in inexpensive equipment, like your iPhone or Android for example can even add 2dB more of grit and saturation on top if this.
No wonder things sound crap.
The Sonnox Pro-Codec can help engineers understand this easier without reading a lot of docs.
But IMHO this problem should be taken a lot more seriously by the artists, mix and mastering engineers and first and foremost the record company.
The cool thing is that with the EBU R128 norm spreading out through broadcast stations today, such mastered songs will truly sound wimpy and weak :-)
Back to the mixer