Hi there. I have a short .ogg with Vorbis stream @56kbps from a 44.1 16bits stereo sample, which weights exactly 135'781 bytes. Packed to ZIP, it goes down to 109'485 bytes, that's a 80% ratio ! Even worse with latest RAR3 archive, with which the file size goes to 104'349 (76%). <sceptical> I admit this sample is a bit repetitive, but I guess the probability of having the beats match exactly the window frame edge is relatively low, so is the one of having multiple similar stream frames. Even though, I don't know if the coding inside Vorbis would let this happen. This is just non-sense to me. Are OGG packets padded and too big ? Or is it Vorbis frames ? </sceptical> So I did the same test with a Lame encoded mp3, with average bitrate at 56kbps too, keeping all frequency range, 44.1 joint-stereo. The results struck me : the source file is 136'028 bytes, which is quite like the ogg reference file. But the ZIP file was 128'287 (94% as opposed to 80%), and the RAR3 didn't do really better with 128'048 bytes (94% too versus 76%). Why do I get such compression ratio with OGG Vorbis ? All files are there : http://gwaingroup.1.free.fr/2003/01/17/ <kidding> It's not a problem for me since my current project uses Ogg Vorbis in a RARed archive, so a better sounding format, which can be packed even more, that's a "twice" better format. </kidding> This could just be done in a somehow better way, am I wrong ? Thanx for any response. <p>Nicolas "Nescafe" Le Gland <p><p><p><p><p><p><p><p><p><p><p><p><p><p><p><p><p>--- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
Nescafe wrote:> Hi there. > > I have a short .ogg with Vorbis stream @56kbps > from a 44.1 16bits stereo sample, which weights > exactly 135'781 bytes. Packed to ZIP, it goes > down to 109'485 bytes, that's a 80% ratio ! > Even worse with latest RAR3 archive, with > which the file size goes to 104'349 (76%).Have you tried rehuff? This is a program that will recompute the huffman codebooks for better packing. Over long files, the default codebooks are usually fairly close to optimal. On short files, rehuff can usually get quite a few percent savings (and sometimes as much as you indicate ZIP/RAR gets). Search the archives for the source code, or wait for the final release if you don't have a compiler. Peter Harris --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
Someone in a post several months ago reported a similar issue. They found they could bzip2 an Ogg Vorbis file and get compression ratios like what you are reporting. Upon further inspection, I don't think anyone was able to reproduce his result, though. Can you tell us what program and options you used to make the Ogg Vorbis file? If there isn't a copyright issue, could you also post the original WAV file somewhere so others could try and see if they get the same result? --- Stan Seibert <p>On Fri, 2003-01-17 at 08:55, Nescafe wrote:> Hi there. > > I have a short .ogg with Vorbis stream @56kbps > from a 44.1 16bits stereo sample, which weights > exactly 135'781 bytes. Packed to ZIP, it goes > down to 109'485 bytes, that's a 80% ratio ! > Even worse with latest RAR3 archive, with > which the file size goes to 104'349 (76%). > > <sceptical> > I admit this sample is a bit repetitive, but I > guess the probability of having the beats match > exactly the window frame edge is relatively > low, so is the one of having multiple similar > stream frames. Even though, I don't know if > the coding inside Vorbis would let this happen. > This is just non-sense to me. Are OGG packets > padded and too big ? Or is it Vorbis frames ? > </sceptical> > > So I did the same test with a Lame encoded > mp3, with average bitrate at 56kbps too, > keeping all frequency range, 44.1 joint-stereo. > The results struck me : the source file is > 136'028 bytes, which is quite like the ogg > reference file. But the ZIP file was 128'287 > (94% as opposed to 80%), and the RAR3 > didn't do really better with 128'048 bytes > (94% too versus 76%). > > Why do I get such compression ratio with > OGG Vorbis ? All files are there : > http://gwaingroup.1.free.fr/2003/01/17/ > > <kidding> > It's not a problem for me since my current > project uses Ogg Vorbis in a RARed > archive, so a better sounding format, which > can be packed even more, that's a "twice" > better format. > </kidding> > > This could just be done in a somehow better > way, am I wrong ? > > Thanx for any response. > > > Nicolas "Nescafe" Le Gland > > > > > > > > > > > > > > > > > > > --- >8 ---- > List archives: http://www.xiph.org/archives/ > Ogg project homepage: http://www.xiph.org/ogg/ > To unsubscribe from this list, send a message to 'vorbis-request@xiph.org' > containing only the word 'unsubscribe' in the body. No subject is needed. > Unsubscribe messages sent to the list will be ignored/filtered. ><p>--- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.