Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Poor Performance Of Md Compression (Read 13060 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Poor Performance Of Md Compression

Reply #25
Yeah, the ATRAC3 specs is more flexible than MP3.. It only specifies the huffman tables, the scaling factors for the quantization module.. but the actual quantization strategy is implementation dependent..

However, I doubted that its huffman codes were even closed to the efficiencies of the MP3 or AAC huffman tables.. Also, the maximum quantized MDCT coefficients that can be coded by the huffman tables are just 31 compared to the escape code technique of AAC / MP3 which allows up to 8191.. That is why there is a need for a separate tone component module to extract the tone information.

There is no specification at all about the psychoacoustic module..

wkw


Poor Performance Of Md Compression

Reply #27
I've been ABX'ing the other MD samples at the PCABX page, my results:

castanets: 16/16  p<0.1%  very easy
trumpets1 :10/10,  p=0.1%  not very difficult if fresh
french:      21/28  p=0.6%
timpani:    21/28,  p=0.6%
violins2:    16/19,  p=0.2%  not too difficult if concentrated
trumpets2:  23/38: p=12.8% did well at the beginning, bad at the end
violins1:  didn't have success at the beginning, so I concentrated on the other clips

Taking into account that I was quite tired and not very concentrated at the end (12 at night here, did all tests simultaneously), I'd say MD compression is not as good as some audiophiles claim it to be.

I'm convinced I can ABX the trumpets2 much more easily, I will try in another moment, when I'm more fresh.

I haven't done same tests with LAME -aps, but I doubt I will get such good ABX results in most samples.

*Note for Joe Bloggs' friends at Head-fi: strange that I can't ABX any audio cable I have at home, but that can ABX quite well most of these MD samples, also when MD is regarded from most people at there as very good sounding and nearly transparent...

 
SimplePortal 1.0.0 RC1 © 2008-2021