I don't know whether this question is inane or not, but I try nonetheless:
In video encoding using codecs such as Xvid or DivX one has the possibility to use 2-pass/multipass encoding, ensuring that the size of the file is (almost) spot on.
Why isn't (and if it is, I'm sorry for bothering you) a similar technique implemented in formats such ass vorbis, mp3 and MPC? Surely you can use vbr algorithms that creates a bitrate and/or filesize that one wants (and I'm not talking about ABR here)? And optimising using multipass encoding should be a good idea?
I'm no programmer myself, I'm just curious about this.
Full Version: Multipass VBR-encoding
As I see it:
1. It is implemented in WMA9PRO and gives VERY good results for 5.1 encoding.
2. It is not implemented anywhere else, becasue:
I. Most of people use encoders in VBR mode, that means that people want constant quality, not target filesize or average bitrate.
II. Target filesile or avg bitrate is needed when you encode on low bitrates (i.e. for streaming or to feat into specific container as CD) but this is not very popular comparing to vanilla music encoding. Thus, 2-pass is not implemented in any other popular encoder.
See this post too to see, that some people do not even understand, why you need any mode except vbr: http://www.hydrogenaudio.org/forums/index....howtopic=20975&
EDIT: grammar
1. It is implemented in WMA9PRO and gives VERY good results for 5.1 encoding.
2. It is not implemented anywhere else, becasue:
I. Most of people use encoders in VBR mode, that means that people want constant quality, not target filesize or average bitrate.
II. Target filesile or avg bitrate is needed when you encode on low bitrates (i.e. for streaming or to feat into specific container as CD) but this is not very popular comparing to vanilla music encoding. Thus, 2-pass is not implemented in any other popular encoder.
See this post too to see, that some people do not even understand, why you need any mode except vbr: http://www.hydrogenaudio.org/forums/index....howtopic=20975&
EDIT: grammar
Thanks guys.
It should be easy, though, to implement something that changes the VBR Quality Settings to better approximate the target bitrate. It would for instance be great for some of rjamorim's tests. In this case you won't need to implement the feature in the codec, you just need a programme that changes the quality settings until the file reaches the target bitrate. I saw some guy proposed this in the other threads.
Could this possibly be done with a .bat-file?
It should be easy, though, to implement something that changes the VBR Quality Settings to better approximate the target bitrate. It would for instance be great for some of rjamorim's tests. In this case you won't need to implement the feature in the codec, you just need a programme that changes the quality settings until the file reaches the target bitrate. I saw some guy proposed this in the other threads.
Could this possibly be done with a .bat-file?
Why don't you want to use ABR if file size prediction is so important?
QUOTE
Why don't you want to use ABR if file size prediction is so important?
Because theorically VBR should be better quality than an equal sized ABR.
In Lame, VBR is flexing bitrate a lot more than ABR.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.