Quote: Yes, perception is very important which is why we make listening tests. You need to eliminate (sub)conscious bias to get valid results. Any audio 'engineer' (sorry for the quotes, but engineer requires an academic degree over here) should know that. Quote: Sure, some encoders indeed can handle above 16 bits, but the resulting MP3 isn't accurate anywhere close down to 24 bits. The only difference between a dithered 16-bit and 24-bit file encoded as MP3 is noise. Low level noise. Quote: I don't understand 'keep the encoder consistent' but yeah your idea to keep as much data as possible is right. Since 320 CBR is the highest possible bitrate (higher than V0) you get the highest possible quality but also much bigger files, and: "but to date, no one has produced ABX test results demonstrating that perceived quality is ever better than the highest VBR profiles" (HA wiki). Quote: Sorry but I don't. Quote: LAME also doesn't filter above 16 kHz with the highest settings. Not even with 192 CBR. If you get a LAME-encoded CBR file with everything above 16 kHz filtered out you got most likely a 128 kbit/s file - something I wouldn't even download for free. Quote: Nice story. Quote: Yeah right, you start believing things for good reasons. Theories are not based on hot air but for example on listening tests.