uncompressed audio confusions

[MDS]

Lossless compression is not perceptual coding like DD/DTS/MP3/WMA et al. It is more like ZIP - the data is shortened but nothing is lost and when it is uncompressed you get back the exact original.

 

[mtrycrafts]

Oh, I agree 100%. I can't hear a difference. When sound level matched, I'm not sure I can even tell a difference between DTS-MA and plain DTS or Dolby TruHD and plain Dolby Digital.
For example, on Batman Begins HDDVD, when I switch back betweem DD and TruHD, I have a hard time hearing the difference that critics have claimed.

I just e-mailed the editor of "The Audio Critic" regarding all these sound formats.
He wrote me back and said, "multi-channel music is over-rated, and all multi-channel formats for movies are GOOD ENOUGH."
I guess that means TAC doesn't care all that much about all these new surround sound formats.

The man behind TAC is getting up in years, don't forget

You need 3 front channels even for music. That was well demonstrated as long ago as the early Bell Labs experiments in the 1920s or maybe the 1930s.

 

[no. 5]

Lossless compression is not perceptual coding like DD/DTS/MP3/WMA et al. It is more like ZIP - the data is shortened but nothing is lost and when it is uncompressed you get back the exact original.

Ahh, so if I may ask, how is that (lossless compression) accomplished?

You have piqued my curiosity.

 

[MDS]

Ahh, so if I may ask, how is that (lossless compression) accomplished?

You have piqued my curiosity.

There are many different algorithms. The basic principle is the same as zip/tar/jar where you want to replace long identical sequences with shorter codes but audio doesn't compress well with standard techniques because there is little redundancy.

The algorithms try to expoit things like the sample at the same point in time for channel 2 differs only slightly from the sample in channel 1 and thus will store the difference between the two rather than the whole value (the difference takes fewer bits than the whole value). Some of them use 'predictive coding' where they calculate a predicted value for the next sample (during encoding) and if it is right, there is no need to store the actual value because the inverse of the prediction calculation will restore the correct value on decoding. Once all the values are calculated they are then compressed even further by replacing the actual values with shorter codes in the same manner that Zip does, except that they use a format called Rice coding whereas text compression routines use some form of Huffman encoding.

I've tried in the past to find actual code for lossless compression algorithms but it must be a closely guarded secret because I have been unsuccessful so far. All I know is the basic principles as outlined above.

 

[no. 5]

Quite interesting, thanks MDS.

 

[kenhoeve]

Don't trust any reviewer over 35. I had an extremely annoying high frequency noise emanating from my ipod in my car stereo, dealer installed, that none of the management could hear. I was so pissed they discounted me that I eventually made them get a young tech to listen, whom immediately heard it. Lesson being that old ears lose high frequencies, quick. And these are the things that lossless is most advantageous for, high db high frequencies are the first to get clipped in compression.

 

[MDS]

All true Mr. Kenhoeve but that loss of high frequency hearing starts at around age 20. Naturally though people are not all the same and will vary.

As for the high frequencies being clipped by lossy compression algorithms that is true as well but once again depends on bit rate. The typical encoder, at 128 kbps, will discard EVERYTHING beyond 16 kHz. Higher bitrates discard less and at rates of 256 kbps, the response tracks the original pretty close all the way to 20 kHz.

The compressed file is not exactly the same of course as the whole purpose of perceptual coding is to reduce the file size by discarding what the codec's model deems to be inaudible.

 

[AcuDefTechGuy]

The man behind TAC is getting up in years, don't forget

You need 3 front channels even for music. That was well demonstrated as long ago as the early Bell Labs experiments in the 1920s or maybe the 1930s.

Yeah, TAC needs someone younger and more enthusiastic running the show!

Here is a question I've wondered about: If your Main Left and Right speakers are A LOT better than your Center and Surround speakers, would you be better off with 2-ch stereo instead of 5-ch music?

If you have 5 identical speakers, then I think 5-ch music is definitely better. But what if your L/R speakers are just so much better in everyway than the rest of your speakers?

 

[mtrycrafts]

Yeah, TAC needs someone younger and more enthusiastic running the show!

Here is a question I've wondered about: If your Main Left and Right speakers are A LOT better than your Center and Surround speakers, would you be better off with 2-ch stereo instead of 5-ch music?

If you have 5 identical speakers, then I think 5-ch music is definitely better. But what if your L/R speakers are just so much better in everyway than the rest of your speakers?

Well, I would only worry about that center and not the surround as that gives space to the listening room, not real direction as in HT. Therefore, you may want to experiment and see how different it is in the front soundstage.

 

[AcuDefTechGuy]

Well, I would only worry about that center and not the surround as that gives space to the listening room, not real direction as in HT. Therefore, you may want to experiment and see how different it is in the front soundstage.

Yeah, I'm going to get the DefTech CLR3000 center channel.