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: How to get rid of joint stereo (Read 49636 times) previous topic - next topic
0 Members and 2 Guests are viewing this topic.

How to get rid of joint stereo

Reply #75 –
Quote from: Mike Giacomelli on
Quote from: MordredKLB on
I know they don't just send raw bits and that the data is losslessly compressed but it still seems to me that when it's uncompressed


Who said anything about uncompressed?  You asked about MP3 and I answered . . .

Quote from: MordredKLB on
either all the samples/data has a standard bit size, or you have to know the bit length.


Of course if the audio is uncompressed every sample has the same size.  Thats why its called uncompressed 

You only get to save bits when you apply some kind of compression.

Wonderslug was discussing M/S as a bit saving measure to simple stereo (which it is in many cases) which wasn't what I was referring to when I used the term compression. It's my understanding that that data is then VLC'd using Huffman to further compress. Maybe I'm completely wrong with regards to MP3 though... I've already stated that I barely even know what Huffman is.

When you start saying things like "The M takes 13 bits to encode and the S requires 9 which is a big savings over simple stereo which would require 28 bits for L+R" there are inherent problems. You can't just have random bit lengths for M/S without knowing how many bits to read for each sample. Just saying it's VLC'd doesn't seem to make the problem go away. When you decode back, you still have the problem of knowing when the different bit sections start and stop. How do you know the M takes 13 and the S takes 9 without either a length field or constant bit lengths (i.e. M is always 15 bits and S always 10)?

How to get rid of joint stereo

Reply #76 –
You split it into blocks and fix the quantiser (bit depth) across the block.

Anyway, with mp3 you're taking about frequency domain coefficients, not samples.


If you don't understand it fully (and why should you?) it makes some sense to trust those that do.

It's a pity people never seem to use that rule in audio - it's like "a little knowledge is a dangerous thing - expect in audio, where a tiny bit of knowledge suddenly makes me an expert!".

Cheers,
David.

How to get rid of joint stereo

Reply #77 –
Quote from: MordredKLB on
When you start saying things like "The M takes 13 bits to encode and the S requires 9 which is a big savings over simple stereo which would require 28 bits for L+R" there are inherent problems.


I know, which is why I told him this!  Maybe reread this part of my post:

"Smaller numbers do take less bits to send then larger numbers, although the exact number used depends on how they're encoded."

You can't compute an exact savings without knowing more, but the general ideas is correct.

Quote from: MordredKLB on
You can't just have random bit lengths for M/S without knowing how many bits to read for each sample. Just saying it's VLC'd doesn't seem to make the problem go away.


Maybe you should look up what variable length coding does instead of just guessing, or if you're not interested in knowing, just take our word for it.

How to get rid of joint stereo

Reply #78 –
Quote from: 2Bdecided on
You split it into blocks and fix the quantiser (bit depth) across the block.

Anyway, with mp3 you're taking about frequency domain coefficients, not samples.

Well I read parts of the link that lvqvl posted last night, but it started making my brain hurt. It did seem that what I was talking about didn't really apply with how mp3 work.

Quote
If you don't understand it fully (and why should you?) it makes some sense to trust those that do.

It's a pity people never seem to use that rule in audio - it's like "a little knowledge is a dangerous thing - expect in audio, where a tiny bit of knowledge suddenly makes me an expert!".

It's not that I don't trust people who understand things better than me, I was asking for more information, precisely because I don't know. A "you're wrong" (which was the impression I got) without explaining (at least well enough for me to understand) why wasn't helping. I'm not some arrogant dumbass here to show you all what idiots you are, I'm trying to figure this stuff out. I like it here precisely because of the knowledge you guys have.
 

How to get rid of joint stereo

Reply #79 –
Quote from: Mike Giacomelli on
I know, which is why I told him this!  Maybe reread this part of my post:

"Smaller numbers do take less bits to send then larger numbers, although the exact number used depends on how they're encoded."

You can't compute an exact savings without knowing more, but the general ideas is correct.

You are correct, I did kind of gloss over that.

Quote
Quote from: MordredKLB on
You can't just have random bit lengths for M/S without knowing how many bits to read for each sample. Just saying it's VLC'd doesn't seem to make the problem go away.


Maybe you should look up what variable length coding does instead of just guessing, or if you're not interested in knowing, just take our word for it.

Maybe reread the part of my post where I said that I did read up on Huffman and VLC.  I'm just having trouble figuring out how it works in this context, which bugs me, which is why I'm asking more questions. Also, "just take our word for it" isn't necessarily the best answer when earlier in this thread cries of "Joint Stereo is Lossless!!!" filled the thread until 2bdecided came along and showed that was a misleading oversimplification for mp3... which was something I didn't know.

Sorry if I've upset anyone... I'm just trying to figure this stuff out. I'll bow out gracefully now and try to figure it out on my own.