Hi, I'm a newbie to audio compression. And I have several question about ogg. 1. How many bits a vorbis sample take ? does the sample size 16 bits ? like the mp3 ? CMIIW 2. How do a decoder recognize how many sample it takes in a second ? does it related with the code books ? 4. Does vorbis use VBR ? 3. What is function of the Windows ? and what is the output of the windows ? Thanks before for your attentions, ____________________________________________________ Yahoo! Sports Rekindle the Rivalries. Sign up for Fantasy Football http://football.fantasysports.yahoo.com
On Thu, Jun 16, 2005 at 05:04:28PM -0700, Mario Antonius wrote:> 1. How many bits a vorbis sample take ? does the > sample size 16 bits ? like the mp3 ? CMIIWVorbis is natively floating point and can in theory represent well over 120 dB of dynamic range. Most people feed it 16 bit-per-channel samples though.> 2. How do a decoder recognize how many sample it takes > in a second ? does it related with the code books ?Yes. The first packet contains the sample rate and number of channels for the data the follows.> 4. Does vorbis use VBR ?Yes, vorbis is natively VBR. The reference encoder has some support for producing "CBR" streams with a more even bitrate.> 3. What is function of the Windows ? and what is the > output of the windows ?The windows reduce transform and lapping artefacts between coded blocks. See the paper referenced from the spec for technical details. Hope that's some help, -r
Hi I am trying to understand the psychoacoustic model used in the Oggvorbis encoder by looking at the source. I found it very difficult to understand. Does any one has any idea on this? Can I get any information on the reference papers (JASA, AES or IEEE ASSP) which can be referred to understand this source code? Can I know who is the author of this module? Regards Venkat
On Fri, Jun 17, 2005 at 03:04:56PM +0530, venkat wrote:> I am trying to understand the psychoacoustic model used in the Oggvorbis > encoder by looking at the source. I found it very difficult to understand. > Does any one has any idea on this? Can I get any information on the > reference papers (JASA, AES or IEEE ASSP) which can be referred to > understand this source code? Can I know who is the author of this module?The code itself is the only documentation I know of of the actual psychoacoustic model implemented, though it is based on a bunch of empirical academic work. Christopher "xiphmonty" Montgomery is the author of the reference encoder. -r
On Thu, Jun 16, 2005 at 05:04:28PM -0700, Mario Antonius wrote:> Hi, I'm a newbie to audio compression. And I have > several question about ogg. > > 1. How many bits a vorbis sample take ?There is no native bit size. Internally, the reference encoder and decoder use floats.> like the mp3 ? CMIIWmp3 also has no native word size. You're proabably just thinking of a specific encoder.> 2. How do a decoder recognize how many sample it takes > in a second ? does it related with the code books ?A deocoder simply turns each packets back into a set number of samples. The samples/second is set in the header.> 4. Does vorbis use VBR ?Vorbis is natively VBR. ABR/BBR/CBR are implemented out of the VBR algorithms.> 3. What is function of the Windows ? and what is the > output of the windows ?Windows are used in short-time spectral analysis manipulation. The simple answer to your question is answered in most undergrad level signals classes. The long answer is still the subject pf PHD theses today. :-) Monty