Hi all,

I've spent the last couple of weeks putting together a Pure Data based streaming radio station and it's finally at the point where I'm happy to let people know about it.

The system is a python script controlling PD and is able to load and fade between patches, the output of which is then streamed over the net. The whole thing is running off a cheap VPS server but it seems to be holding up well.

You can listen to it currently by going to http://radio.rumblesan.com and there are more details available on the technical stuff at my blog, rumblesan.com

One of the things I'm hoping is that I can get some other people involved in this. I have lots of ideas and things I'd like to add in future, but the first thing that really needs doing is making more patches. Lots more patches!

Over the weekend I'm going to write up how the control and interface stuff works and all the info needed for people to create patches. All the code and the patches made so far are up on GitHub if people want to have a look, links are in the blog post.

This is really great. I am already thinking up patches

I keep losing connection. It plays for 30 seconds and then goes away for 10 mins and then plays for 30 seconds and drops or it might not come back at all. If I reload the page or press stop and then play it will play for 30 seconds and then lose connection again.

Can I use PD Extended for the patches?

hhmm, I see what you mean about the connection issues, i think it might be loading issues on the server finally rearing their head I'll look into it and try to get everything running smoothly again. It was all running fine earlier this afternoon so might be

Currently the only libs i have loaded are maxlib and pdogg. I'm happy to add more libs but wanted to try and keep things minimal to begin with.

glad to hear you're interested tho

EDIT: I've restarted the process and it seems to be fine again now. Maybe i'm doing something stupid and leaking memory somewhere. I'll try and get it fixed

I've now written up a post on the technical side of the radio. What it's doing, how it works and how to write patches for it. all the info is at http://www.rumblesan.com/?p=267

let me know if you want to help out.

btw audiolemon, I've now had the stream running for two days and no issues. i suspect that the problem is due to someone else on the VPS cluster doing something cpu intensive and it affecting my server (total speculation in truth). I'm going to keep it here for the moment but i might look into a cheap dedi server soon.

This is a badass idea! Thanks for making it.

I really really like the idea, and also the patch currently running is not bad! Sounds like Formula 1 cars passing by
If you have technical/bandwidth problems, post back here - I'm sure the idea is worthy to be hosted by some tech university if presented in a good manner. This could inspire lots of people to write generative music patches.

BTW, running with no problems for 10 minutes now.

It worked with out issue each time I visited, interesting project, I've downloaded all the files now, may well have a crack at writing something.

glad that people thing it's a cool idea. i've got so many ideas for it but it really is a matter of prioritising what needs doing right now, and more patches is the important thing.
thankfully the five i've got on there at the moment seem to be interesting enough.

danijel, i hadn't thought about asking in at universities for help but it's not a bad idea i suppose. currently i'm fine with operating it out of my own pocket since it's very little money and the VPS makes it easy to reinstall OSes incase i balls things up.

lead, i've got a few changes to the master patch that i need to drop into the repository but they don't change the functionality, just clean the patch up a bit and make things much neater. i'll push them in tonight. also, there are a couple of the patches that have some mistakes in, delay objects all need to have $0 arguments to make sure they're unique and i haven't done that with some of them

I've got some changes to the master patch that i need to push to the github repository but they don't change the functionality, just make things much neater.

You could try the SANlist http://lists.sonicartsnetwork.org/listinfo.cgi/sanlist-sonicartsnetwork.org looking for someone to host, plenty of academics that might be interested on there.
Nice one on the updates, will download when I get a chance.

OK, changes are in the GitHub repositories. The master patch has been neatened up and the radio patches have had their delays, sends and receives made unique so no interference problems hopefully.

The patch attached was inspired by hamster's "Meditation machine" approach (extended phase relationships driving random parameter value selection?), plus a couple of its nuts and bolts, plus a self sampling module.

I included the original too sans $0 and PatchCom for reference.

Hope it works okay...

http://www.pdpatchrepo.info/hurleur/lead_asap.zip

Nice one, i'll have a look at it when i get home later and, assuming it's all fine, i'll put it into the repository and roll it out to the server tonight. it sounds pretty interesting.

Is there a readme in the archive with your details/info/etc? I want to try and make sure that people get credited for the patches they make and add.
One of the things i'd like to have later on is a way for people to browse the patches that are on the radio and see who made what. the first step with that is probably getting the meta data for the stream updated when a new patch is loaded so it displays the creator and patch name.

I'm having a look at the patch now and i'm really impressed. it sounds really good and is exactly the kind of thing i want to have on patchwerk radio.

i've looked at the version you've modified to work on the server and the only issue I can see are a couple send and receives within the hann-window and fft subpatches dont have $0 based names which i've now fixed.
there was also a loadbang in each of the fft subpatches that seemed to be turning the dsp on, setting a local blocksize and sending the same value to a single receive in the hann-window subpatch. i've changed it so that blocksize/overlap is set with the block~ arguments and the hann sample size is set by a loadbang. I checked with the original version you included and the changes seemed to be fine, i can't see why they wouldn't but let me know if i missed something.

I'll put the patch into the repository with the name ASAP, because that's what you've called it here heh. it looks to be a bit heavier on the system resources than any of mine so far so i'll be interested to see how the server copes with it.

also i'll turn on the "extra" lib so that expr~ and kalashnikov will work. i think they were the only extras you used but again, let me know if i missed something

Got it working but i've managed to break something else..... and i;m not sure what. Have taken it offline for a bit whilst i fix the problem

Well that took longer to fix than i anticipated, either way, everything is back up and running and Lead, your patch is now in the pool.

I've added it into the patches repository, fixed the above changes i mentioned and added a short info file.

seems the server handled it pretty well, so i don't think i need to worry so much about loading in future.

Great stuff, caught it on there this morning.
Forgotten I'd left those objects in, was trying to keep it vanilla, oops...
Checked your changes and everything is running fine as far as i'm concerned- nice one!

Here's another patch attached, will shoot you an email in a bit to say hi and give you a url for the info.txt

http://www.pdpatchrepo.info/hurleur/lead_tonnes.zip

what a great idea!

I'd like to submit a patch - my minimalchine - too, but there are some [counter] objects from pd-extended/extra/cyclone. Is that a problem? The patch was posted here earlier this day: http://puredata.hurleur.com/sujet-5158-minimalchine if you like to have a look.

I looked at the other patches in github and now i'm thinking:"I just need to exchange the [dac~] with [patchComs](?) and add 2 more [loadbang]s on the volume and the on/off Trigger". Is that correct?

Hi Az, thanks for the interest. i'm currently at work so can't have a look at the patch but i'll check it out later this evening and let you know if it needs any more changes.

usually it is just a question of replacing the dac with the patchComs abstraction and making sure that any sends, receives, tables etc are unique to the patch.

No worries with the counter object, i turned on the cyclone and ext13 libs for lead's first patch. i'm happy to turn on more libs if people need them. i probably need to have a look and see if there's anything too dangerous or exploitable really. i don't think anybody here would deliberately do that but it's still a good idea to check.

Well, there are some [s~ ch1]... i could get rid of them, since they are not really needed. Maybe even get rid of all the r/s objects...

If you're concerned about security, then i would just change the owner of Pd-Process and the according permissions of the Webserver and you should be safe from data manipulation.

hey az, i've literally just gotten around to having a look at the patch, spent the evening wrangling with code and setting up a nice way to run up a test server for checking changes i make.

i really like it and i'd love to have it running on the server. the sends and receives seem to be the only problem currently. you might want to get rid of the Mix table as well since nobody can see the GUI on the server heh.

you said in the other post that you wanted to spend a bit more time improving and bugfixing it. if you want to do that first and then let me know when you're ready to have it go up then that's fine, but i'm happy to take it now and then you can send me an updated version later.

currently the python/pd stuff is all running under it's own user so that's already covered. it's more that i dont know what else is in the libraries but you make a good point that that should be enough to keep things secure.

Lead, i'm afraid that patch is a bit too CPU intensive for my poor server heh. it will play a bit but then keeps dropping out every 5 or 10 seconds. It's a pity as i really like it, makes some amazing sounds. perhaps i should invest in that dedi sooner rather than later

Ah yeah got a bit carried away with oscillators there, here's a smaller less hungry version. Kind of spoils the effect, but if it fixes things your end I think I need to modify my design approach (massed granular coincidence? heh) a bit.

http://www.pdpatchrepo.info/hurleur/lead_tonnes_redux.zip