You know, we really need a streaming FAQ. There seem to be LOTS of undocumented command-line switches for oggenc that help in limiting bitrates that folks don't know about. The recent thread with Derek at CD Baby brought out a lot of good info that I didn't know, and I consider myself fairly plugged-in. It would be helpful to have something like: Q: How do I limit the bitrate to absolutely, positively no more than X kbps, even for a millisecond, starting with source material at 44.1 KHz/16 bit stereo? A: Try something like: oggenc --frob=X --managed --downmix --bar=FOO *.wav Q: Okay, that worked, but I've decided I can live with the occasional millisecond of data over X kbps. How can I improve the quality? A: In that case, you should be able to considerably improve quality with something like the following: oggenc --quality=INSANELYGREAT --shoot_for=X *.wav Is there anything like this out there already, or should someone write one? -- Graham Mitchell - computer science teacher, Leander High School "Never take a job you wouldn't do for free." --- >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-dev-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.
Let's move this to vorbis@ where it belongs. Graham Mitchell wrote:> You know, we really need a streaming FAQ. There seem to be LOTS of > undocumented command-line switches for oggenc that help in limiting bitrates > that folks don't know about.Actually, all the useful command line switches are documented. :)> The recent thread with Derek at CD Baby brought out a lot of good info that I > didn't know, and I consider myself fairly plugged-in. It would be helpful to > have something like:[...]> Q: Okay, that worked, but I've decided I can live with the occasional > millisecond of data over X kbps. How can I improve the quality?The thing is that 99.9% *can* live with the occasional bandwidth peak - that's what client-side buffers are for. It'd really suggest taking the other approach of telling people how they can optimize their results with the documented features, instead of telling them first how to break Vorbis. There aren't many applications and devices that a typical consumer sees that require absolute CBR, and streaming (even at low bitrates) doesn't belong to those. If this FAQ contains a question "I want to use Vorbis on <device> and its chipset/whatever requires absolute CBR - it has a tolerance of n milliseconds, how can I achieve that?", then the first part of the answer must be "Note that Vorbis wasn't designed to do this.". I'd rather be honest and helpful by saying "<x> does a much better job, use <x>" than make people hate Vorbis because they tried it in an enviroment where it's not meant to work. I just tried the advanced option "bitrate_average_window" and set it to both 0.05 and 0.001 seconds, and used both -m and -M. While my test bitrate of 96kbps was never exceeded in my test, it dropped as low as 88kbps according to Winamp in one place. :P This means that either -m doesn't work properly, or the bitrate_average_window doesn't quite work for the minimal bitrate. Odd enough, by using this option I greatly increased the overhead - the smaller the window, the bigger the resulting file. I don't know whether this is due to Ogg or Vorbis... maybe someone can shed some light on this. If it's Ogg, libogg2 or a completely different transport might be able to fix this. Ah well, all I mean is that all this shows that Vorbis is not designed for CBR and those advanced options are undocumented for a reason. Together with the bitrate management engine (assuming it's not being abused with advanced options), the current implementation of Vorbis allows its use in a huge majority of possible cases. Except for a few, and it doesn't *have* to work there, imo. I suggest, to the one who has the time to write the FAQ, that only documented features are being discussed. There are many people out there who think they're doing something extremely cool by messing up Vorbis' great default settings. That's also the reason why many lame encoded MP3s sound crappy, although --alt-preset provides excellent results. People always try to look smarter than they are. :) <p>Moritz --- >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.
Graham Mitchell wrote:> > Q: How do I limit the bitrate to absolutely, positively no more than X kbps, > even for a millisecond, starting with source material at 44.1 KHz/16 bit > stereo?There is no such thing as instantaneous bitrate; bitrate is an average number of bits over a certain time period. Compare with "temperature". <p>Segher <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-dev-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.
Hi: I guess this raises another question. Should there be access to such exotic features of libvorbis in ices? I mean, ices is meant to be the tool of choice and all that for broadcasting. hmmm. maybe this should be on the icecast(-dev) mailing list. Geoff. <p> -- Geoff Shang <gshang@uq.net.au> ICQ number 43634701 Make sure your E-mail can be read by everyone! http://www.betips.net/etc/evilmail.html Please avoid sending me Word or PowerPoint attachments. See http://www.fsf.org/philosophy/no-word-attachments.html <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-dev-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.
On Thursday 30 January 2003 21:44, Geoff Shang wrote:> Hi: > > I guess this raises another question. Should there be access to such > exotic features of libvorbis in ices? I mean, ices is meant to be the tool > of choice and all that for broadcasting. hmmm. maybe this should be on the > icecast(-dev) mailing list. > > Geoff.Yes, there should. It's just boring to add, so I haven't. Patches welcome. Mike <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-dev-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.