Slashdot Mirror

I've been playing around with PeerCast lately and it's really great although it has some major shortcomings. For example if you have dynamic IP the channel ID will change every time you get a new IP, you can't have multiple sources feeding the same channel (mirrored Shoutcast/Peercast stations or simply a redundant feeder), and you can't get the stream from multiple sources if one of them don't have the bandwidth to send the complete stream.

What is needed is a PeerCast-like service with support for multicast and stream-rebuilding from multiple sources and something that identifies the channels by a public key instead of building the ID by making a hash from a number of variables on your system (including your own IP and the IP of the feed).

Unfortunately, this - and esoteric p2p networks - are literally the only things that use multicast, and it's a complete fucking bitch to set up. Support on the streamer end is there - WMS (yes, Windows Media [retch]) does pretty well - but I've never, ever seen it used in favour of UDP unicast.

The future might look a bit more like Peercast, integrated into some sort of massively deployed p2p network. Only with a stream-to-disk. And the bandwidth will, I'm afraid, have to be kept down - use Vorbis and minimise the losses...

Yupp and guess what it alread exists!

Peercast

...by giles, the author of Peercast Click here to get peercast
You also need NSV video plugin for Winamp Click here to get Winamp

Something like Peercast, perhaps?

Sounds like what you want is Peercast. You can't split up files the way bittorrent does for broadcasting, so what it does is allow each client to also broadcast. I haven't used it much as I'm on a metered internet connection but it seems to work as long as it doesn't matter to you that some clients will get their data later than others.

Their servers would die because of the overload

Not necessarily. Unless I'm misinformed IP broadcasting enables streaming to multiple recipients using one TCP-feed. And solutions like PeerCast are creating yet another option.

Sounds interesting but I'm not going to install the Crossover plugin just to see it. Is it possible to view this with Mplayer now that it has full support for QuickTime?

P2P Radio seems to be working really well and I'm sure it's the same for video too. The RIAA is going to have a few headaches over these too... first they had unstoppable file-sharing and now they'll have same for live video streams and radio. :)

Have you ever heard Epiphany Radio? It's a shoutcast station I used to listen to, until I ran into a 12 user limit imposed because the broadcasters couldn't afford the bandwidth to support many users. Thanks to peer to peer technology (the Gnutella protocol), I can once again listen to this station, via their peercast stream.

Have you ever heard Epiphany Radio? It's a shoutcast station I used to listen to, until I ran into a 12 user limit imposed because the broadcasters couldn't afford the bandwidth to support many users. Thanks to peer to peer technology (the Gnutella protocol), I can once again listen to this station, via their peercast stream.

I thought of this already, as soon as peeling was mentioned as a possible feature.

This is the real reason Vorbis could rule supreme over all other codecs (well, more than it does, what with winning the listening tests[1], and being both patent-free and open-source as well).

What peeling means: Adaptive streaming.

Because you can reconstitute the peeled segments, you can buffer extremely low quality stuff well in advance, increasing the depth of the segment (and thus quality) as the segment gets closer to playing. Use a UDP protocol in such as way as to measure the bitrate coming through and adapt more closely to changes in the bandwidth as, say, the modem user uses the web. Obviously, make sure the low quality stuff gets through first [3]. Where even the best streaming stuff (even WMP9-rc, which cannot do this for the same reason mp3 cannot) just cuts out because of insufficient bandwidth, this would just lower the quality a bit - and with Vorbis' quality/bandwidth ratio, you might not even notice. 128kbps nominal (q3) Vorbis 1.0 is very acceptable and to many users indistinguishable from CD, and you could potentially broadcast high quality 160kbps streams or so (that only the real golden ears can tell apart from the source, except in really extreme cases).

Add this into the basic protocol of a future version of, say, Peercast and you would have a very, very sweet 'net radio system, potentially dramatically more stable and scalable than anything currently existing.

Also, handily, this could be used to listen to a file while downloading it from a peer-to-peer network. "Holes" could be left in the file as it's streamed using the above manner, with the rest filled with the pieces of stream that made it. The rest, if there is any rest (and thanks to network conditions, there may well be), comes over afterwards, possibly using idle bandwidth. Now you can download a -q8 Ogg Vorbis audio file, indisputedly indistinguishable from a CD, and listen to what you are downloading at the same time (possibly to check it is what it says[1]). On a 56K modem. Sweet.

[1] In everything but classical music, on which mp3pro did better - but frankly, the kind of person who listens to classical music will want to use high bitrates or FLAC, no?

[2] A note to future peer-to-peer designers - Learn from Freenet and the warez scene - keep downloads and uploads anonymous, but digitally sign uploaded "releases", over time these groups who release will become trusted for high quality - presto, the p2p quality problem solved, and the attack of seeding networks with duff files fails because the Judases can't fake a release from a known group.

[3] Imagine a graph. Time axis on horizontal, with a line relentlessly moving to the right (time index of frame currently being played, naturally). Quality on vertical, ranging from 0, which is a skip, smoothly up to 1, which is the full quality of the source Ogg Vorbis stream (unpeeled). You have a given, initially unknown amount of bandwidth which will fluctuate over time - small fluctuations continuously, with big fluctuations from time to time as other applications (e.g., web) inevitably clog the connection. Your mission, should you choose to accept it, is to design an algorithm which uses the bandwidth to fill the graph vertically as much as you can, while always staying ahead of the line despite normal little fluctuations or hiccups in bandwidth, and ideally even in the event of the network getting a bit claustrophobic. Ideas, anyone?

Ideas and methods mentioned in this posting are hereby placed in the public domain. This posting, and the mass public witness of it here, may act as prior art to invalidate any future patents based on these ideas or methods.

(And anyway, aminorex already mentioned the idea basically in #4808308, I'm just expanding on it.)

"The underlying claim from your argument is that P2P solves a technological problem - namely, bandwidth limitation."

"You are trying to justify a technology by mating it to a perceived, likely non-existent problem or future benefit of indeterminate nature."

"The underlying claim from your argument is that P2P solves a technological problem - namely, bandwidth limitation."

"You are trying to justify a technology by mating it to a perceived, likely non-existent problem or future benefit of indeterminate nature."

In the United States (at least), everything made since 1923 was, has been, and still are copyrighted, even if they were never registered with the copyright office. So everything you see on a peer-to-peer network is indeed copyrighted.

A more approriate question (as some of the responders have answered) is if anyone has used a peer to peer network for a legitamite purpose. The problem here is that the issues are quite grey. If I have Game X, or Game System version 1.1 can I download copies of the games/BIOS/etc. online for use with emulators/replacements for broken discs/etc.? If an online broadcaster, paying royalty fees, uses ABAcast or Peercast to distribute their works, do I in turn have to pay royalty fees since I am rebroadcasting them?

Unfortunately, there is a major gap between what people think they can do under copyright law and what they actually can do. While I have not extensively researched the above (IANAL), technically, all the above commonly considered legitamite things are *illegal* unless you have worked some deal out to repay the copyright holders.

The problem you really should be asking is if anyone uses P2P networks to delibrately distribute their copyrighted works, either as a primary or secondary channel. A few minor bands likely do. The next question is if you'll ever find them on Slashdot. And I do not know the answer for that.

Note I personally have *never* used Napster, Gnutella, Kazza, or any of the other networks, mainly because being caught doing so may jepordize my ability to be hired in certain areas. I used to be one of those nasty college network administrators trying to keep your P2P usage down because it overloading our bandwidth, and we could not order a significantly bigger pipe because our local phone switch could not handle it. Feel free to flame me for my ignorance as you will.

After reading through some of the message boards on their homepage, it seems the questions of it being released under a GPL type license were not met with too open of arms.

Interesting thread to read here. It will be interesting to see if this works or is just a stunt to pull in a bigger listener base to increase the network size. After all from the looks of that license, it seems to leave an out for everyone else to do the coding and them to sell it as a commercial product.

Is it like icecast?

Independant artist could setup a custom web page on some cheap or free host, use paypal-like for payement and setup a peercast stream on a 56Ko or DSL connection. Nothing fancy but not so hard to setup and really independant. The peercast stream could contains URL to the artist's homepage.

There are several P2P streaming solutions in development. My personal favourite is:

http://www.peercast.org/

It works quite well, considering it's only at version 0.116B, the developers say that it's going to be open sourced, and there is both a Windows and Linux version available now (Mac on the way).

Not a lot of music available yet, unless you're into Japanese Pop, computer game tunes, or speeches and debates from the EFF (the latter are quite good listening actually, keep it up whoever you are).

Each stream can have basic info about the "station" and a link to a web site. It works quite well on my 56K modem.

Skev

Peercast
I'm surprised noone has mentioned Peercast as a viable way of broadcasting music for free...I think that's one of the reasons that the peercast programmer made the program...

The death of Napster-style centralized p2p lead to the dominance of Kazaa-style distributed p2p, and the death of traditional streaming will lead to the dominance of distributed p2p streaming.

Please take some time to write the RIAA and thank them for their support in advancing the state of the art in free content distribution. :^)

And the lastest version supports easy relaying of shoutcast streams :) Peercast.Org

From the peercast website (emphasis mine):

Our aim is to make the PeerCast client code open source (GPL) at some point in the near future so that it can be checked and ported to as many platforms as possible. Our main priority now though is getting the system stable and usable. One reason for not releasing the source code so far is literally because we haven`t had the time yet.

If they want to GPL the code anyway, why not do it sooner rather than later?

They have

1. Download the PeerCast (Linux or Windows) client.

2. Download Icecast/Darkice/Oddcast or any other OGG/MP3 broadcaster. (You don`t need the server part of Icecast, just the "Ices" part)

3. Decide your playlist, name and bitrate then point the broadcaster at your PeerCast client.

No extra costs, no extra hardware, nothing more than your current Internet connection. Its P2P which means all of your listeners share your broadcasting bandwidth, its anonymous, free and frankly, rather good.

You can even use it to replace an Ice/Shout server if you have enough bandwidth to serve all of your listeners.

There are at least two peer to peer Internet radio networks: Streamer
and PeerCast.

You could also serve hot streaming mp3s with this very cool perl script

Or you could just bliss out to some great trance music ... as long as they can afford to keep going.

Eponymous Mallard
"If it quacks like a duck, it's the Eponymous Mallard"

You can stream OGG via peercast and either oddcast/winamp or ices2 on linux.

The quality is brilliant, even for a 24kbps ogg stream!

There's always peercast, its looking really good recently, runs on windows and linux, supports Ogg Vorbis and has a ton of free radio stations out there, growing daily.

check it out, who needs all this paid subscriber rubbish.

I hope more artists sign up for this.

Hell, I don't want my favorite artists distracting themselves with this business bullshit, unless being a fighter is part of what they already are, like Ani diFranco or Courtney Love. Life is too short, time too scarce for the few genuinely talented artists we have to go running off on tangents.

This is a battle that we, the consumers, should be fighting. If we decide, en masse, not to play the RIAA's game, what the Hell can they do.

Can I suggest that PeerCast (as discussed on /. earlier) is a very good place to start.

And, remember, if we really want to stop these bastards shagging us, we must always remember that our participation in P2P has to be about growing a new, fairer system, not just getting our hands on free stuff.

There's apparently a winamp plugin in the pipeline.

If this could get bundled with the regular winamp download, I think we'd be on to something.

I think the folks at winamp would be interested in doing this; it's an interesting 'selling' point. Download this player, get instant access to millions (?) of songs instantly and without further downloads.

The gnutella network, if I read things correctly, would benefit from the incremental bandwidth of Joe Sixpack and his brethren.

Win-win situation?