Talk:Wikimedia Radio

From Meta, a Wikimedia project coordination wiki

Previous discussion can be found on the talk page of the draft proposal on Wikinews (See n:WN:RADIO). --Brian McNeil / talk 09:33, 28 April 2008 (UTC)[reply]

Possible contributors[edit]

If you are interested in being involved, please sign below. Brian McNeil is trying to reactivate this as a follow-on from doing WikiVoices on December 23, 2009.

Required development[edit]

Due to a streaming service using software that the WMF does not currently employ, some development work is required.

Free software exists to serve up the stream (Eg Icecast), and another package, Liquidsoap, can feed content into IceCast, but it is far from user friendly.

Thus, either a MediaWiki extension to drive Liquidsoap is required, or some form of bot that can read a schedule from a wiki and pass it into Liquidsoap. --Brian McNeil / talk 09:46, 28 April 2008 (UTC)[reply]

This could easily be achieved (I believe, anyway) by using the python bindings for liquidsoap in conjunction with the pywikipediabot framework - would probably need someone more experienced in Python than me to do it but i've had help from the devs of pywikipediabot before now - you can probably catch them in IRC. --Skenmy 10:47, 28 April 2008 (UTC)[reply]
have you spoken with ST47? 62.56.54.110 14:51, 28 April 2008 (UTC)[reply]
I'm not sufficiently familiar with python to do this, but when I was using IceS, which is another option than Liquidsoap (which I have not looked at yet) I was able to update the schedule regularly by running a perl script, which looked at the wiki, which had a playlist of an hour's worth of content, loaded that, converted it from user-friendly show names into the data IceS needed (which is just a path name), and downloaded the audio (i.e., from commons). It's also possible to, when loading this content, load copyright data from the page and use that to decide whether to play a nice message beforehand, which would be prerecorded for each contributor/license. An example schedule could be something like this:
  • Wikipedia Weekly episode 20
  • Audio enwiki signpost april 27
and would be set up by the project's admins and reloaded every couple of hours. A script would know how to convert from "Wikipedia Weekly episode 20" to find where WW episodes are stored, download #20, get the license info, and add it to the stream in the appropriate order. ST47 18:42, 28 April 2008 (UTC)[reply]

This idea needs a 'test-wiki' ShakespeareFan00 00:30, 29 April 2008 (UTC)[reply]

Let's bug brion to make us one. ST47 10:27, 29 April 2008 (UTC)[reply]
I'm all for getting Brion involved, but with SUL and Flagged Revisions featuring large on the horizon I think his input may be restricted to giving guidance on how to get this picked up by Google's "Summer of Code" people and an install and run from a wiki radio service developed. Between now and then there is a lot to clear up about how we are going to get this thing started and build momentum. --Brian McNeil / talk 12:36, 30 April 2008 (UTC)[reply]

I managed to get icecast to stream a list of media files from my personal wiki useing IceS and a shell script. user:Bawolff 05:08, 11 May 2008 (UTC)[reply]

Wild-Assed Content ideas[edit]

It doesn't matter what project you have the most edits on, there's likely some way you can think of packaging some of the material you've created or worked on in an audio format. Most of the low-hanging fruit is already listed for each project on the content page, but there's always new ways content can be generated that could be broadcast. Have an interest in the really strange? Start recording your own "Weird of Wikipedia" series as usable content.

A very significant section of the work involved in making this project take off does not involve sitting behind a microphone. Anyone who has given a presentation more than five minutes in length will tell you, writing material to be read, or to be read out loud, are two entirely different processes. We need script writers as well as the dulcet tones of our readers and announcers. We need wizard wordsmiths who can come up with taglines and intros for regular segments that people will quickly remember; and, although an audio medium, we need people who understand iconic aspects of any endeavour like this. There will be certain people are too heavily accented to be doing the audio work, and there will be those we'd want to do everything. With one of the comparisons of my project outline being an online version of BBC Radio 4 I think we'll be looking for the people closest to "BBC English" or, from the other side of the pond, "CNN English". For a great number of people who have English as a second or third language there are sounds they cannot pronounce because they have no equivalent in their native tongue.

So based on some people potentially being excluded from being 'on air', one of the things I'd like to see with WN:RADIO is credit above and beyond the CC-BY we work to on Wikinews. This will not always be possible, but if some team of people has done a 30 minute segment on a composer using his WP article and his most well-known works I think that as many people as possible should get mentioned in the streaming app's ticker. (Aside, is there a standard term for this?) --Brian McNeil / talk 13:05, 30 April 2008 (UTC)[reply]

Crazy suggestions[edit]

If you have any crazy content suggestions list them below. Ideally things where you can do a significant part of the work without much help, but if you think you've hit upon a sure-fire winner list that too. --Brian McNeil / talk 13:07, 30 April 2008 (UTC)[reply]

Former Wikicast Ideas[edit]

  • 'Netural Zone' - Points of View/Right to Reply/Mediawtahc type programme.
  • 'World Trade Bulletin' - Business news , and prices assuming there are free sources for the data
  • "S'unosi" - Sciecne and Technology issues (Possible sources are Wikipedia/Wikiversity)
  • "Whitehall1313" - Radio Drama in the Twilight Zone vien - Possibly out of scope for this project
  • "The Lincolnshire Poacher" - Radio Drama - Possibly out of scope for this project
  • "This week in Wiki" - Redundnat as WikipediaWeekly/NotTheWikipediaWeekly covers this
  • "On the Day" - A list of dates or connnections about the current, or forthcoming dates...
  • "Signpost" - A modifed version of the signpost for Radio..
  • "This Week in Amatuear Radio" - External item- was being considered but currently the external source concerned has incompatible licensing.
  • "Engineering Bulletins" - Technical stuff relating to streaming, Vorbis, audio recording.
  • Test Transmission material -( See Commons category)
  • "The Arabian Nights" - Readings from this work, Ideally requires someone famaliar with arab culture, and a good oral storyteller...
  • "The Desk" - A Selection of questions from the Wikipedia Reference Desk, might need filterinf in respect of broadcast guidelines though
  • "Debate" - The proposal came out of suggestion to record and air semi-live debates held at certian

locations ( for example the Cambridge or Oxford Unions.) assuming the availabilty and acceptance of specfic speakers.

  • A News digest for the ANZAC region,
  • "How to wreck a House!" - The title was a partial joke, as the intention was to provide with the assistance of a Wikibooks D-I-Y guide, helpful hint on how to do some work properly.
  • Weather Forecasts - Problem is source data, timeliness and scripts...
  • "Your Life.." - A programme covering social issues (possiblity for Wikiversity involvment),

topics under consideration were

    • "Your life.. Up in smoke.." - On smoking issues.
    • "Your life.. At breaking point.." - On stress.
    • "Your life.. off the waggon.." - On alcohol.
    • "Your life.. In the minority.." - On ethnicity and related issues.
    • "Your life.. On public view.." - Data Protection & Privacy issues.
  • "Over the Wire.." - Historical documentary about notable 'escapes'..
  • "A Walk in..."

ShakespeareFan00 15:04, 30 April 2008 (UTC)[reply]

ConMan's Crazy Contribution[edit]

  • Kind of based on ShakespeareFan00's list above - some audio documentaries. Somewhere between Wikipedia (since it will be on a single topic or group of related topics) and Wikinews (because of the allowance for some original reporting in the form of interviews and the like).
  • How many Wikinews reporters have recorded interviews? I know David Shankbone's got a fair few, and if they're of a decent quality it would be worth playing a few of them as special features.
  • Surely there are a few audio recordings now in the public domain, that perhaps other broadcasters might have missed? I'm thinking 70-year-old radio plays and comedy acts, and maybe some pre-radio recordings as well. Some of the external links in w:Old-time radio and w:Audio theatre might prove fruitful.

Confusing Manifestation 04:25, 1 May 2008 (UTC)[reply]

The issue with Old Time Radio, is the 95 year copyright term, I don't honestly think there was

broadcast radio 100 years ago... ShakespeareFan00 10:13, 1 May 2008 (UTC)[reply]

Seriously 95 years? How did they score that? Still, I'm pretty sure Thomas Edison's recording of "Mary Had a Little Lamb" is out of copyright now, right? ;) Confusing Manifestation 13:07, 1 May 2008 (UTC)[reply]
In respect of the US the last Copyright Extension Act IIRC, it made "works for hire" which a lot of OTR would be, have a 95 year term.. In respect of UK works, it could be longer given that (C) in joint works lasts the lifetime of the longest surviving contributor + 70 years.. 62.56.117.55 16:32, 1 May 2008 (UTC)[reply]
Actually, if I'm reading w:Copyright, Designs and Patents Act 1988#Duration of copyright correctly, sound recordings and broadcasts last a straight 70 years in the UK. Of course if it's a recording of a piece of music, then the composition itself is life+70, but if it were a spoken word recording, or a recording of a piece already in the public domain, it looks like anything from before 1938 is now fair game (in UK law, at least, and only to my non-legally-trained eyes). Confusing Manifestation 00:51, 2 May 2008 (UTC)[reply]
Well, It's the US 95 year term that matters, as the US doesn't recognise shorter term, and

its where the WMF servers are.. ShakespeareFan00 10:55, 2 May 2008 (UTC)[reply]

Yep. Shame. Of course, the 95 years just happens to cover all existing broadcast radio (which started around 90 years ago). What about stuff that had fallen into public domain before the Copyright Extension Act (i.e. stuff between about 1918 and 1923)? And what about 100-year-old phonographs? Confusing Manifestation 01:17, 5 May 2008 (UTC)[reply]
Public domain phonographs, hmm... Depends on the work they contain..62.56.117.55 12:19, 5 May 2008 (UTC)[reply]
In fact, the Australian National Film and Sound Archive would appear to have a number of recordings from the turn of the 20th century, which by my understanding would be PD in most countries. Some of them have even been made available in wav/mp3 formats, so it would appear to be possible to use a few of those. Confusing Manifestation 01:35, 5 May 2008 (UTC)[reply]

In the USA sound recording before 1973? are under state law are varies by state. Zginder

Is this legal?[edit]

Didn't this come up on the foundation-l mailing list and was informed that the GFDL, does not allow for public proforance? Zginder

Enquiries into that are being made. --Brian McNeil / talk 06:16, 6 May 2008 (UTC)[reply]
Does OGG Vorbis allow for embedded license text? If so there maybe some 'creative' ways of

meeting GFDL requirments, according to someone deeply involved in the 'free' software world. 62.56.117.55 15:42, 6 May 2008 (UTC)[reply]

Where does it state that we are broadcasting under GFDL? I was under the impression that this would be CC-BY? --Skenmy 16:30, 6 May 2008 (UTC)[reply]
CC-BY would only work for Wikinews and nothing else. Zginder
Not necessarily. The print edition includes some limited gfdl content and we have other licensed images. The main news stuff would have to be cc-by I think, but theres nothing stopping us from using some wikipedia content as while (minus the gfdl and broadcast thing). (and ogg allows some meta data. thats what the man page of ffmpeg2theora suggests anyways, as well the program exiftool can read ogg meta data supposedly). 24.65.78.101 20:13, 9 May 2008 (UTC)[reply]

(outdenting) Hmm... IANAL, but metadata sounds promising. Xiph.org has some documentation available at http://www.xiph.org/vorbis/doc/v-comment.html, and they specifically suggest "license" as a comment field. We may not have enough space to cram the entire GFDL in, but there's plenty of room to say "Licensed under the terms of the GNU Free Documentation License, available at blahblahblah". Additionally, the suggested field names include both "artist" and "performer"; following their suggestions we could link "artist" to the article history and use "performer" to indicate the person or persons reading the GFDL'd text. Finally, the suggested "organization" field would be a good place to plug the relevant project.

Of course, I'd want to run this by Mike Godwin first, but I do like the way this is shaping up. - jredmond 03:36, 14 May 2008 (UTC)[reply]

Creative Commons collaboration[edit]

Mirko Linder of CC Sweden got in touch with me, apparently they're already well on the way to getting something up and running. A mailing list has been started, go here to subscribe. --Brian McNeil / talk 06:16, 6 May 2008 (UTC)[reply]

This is now operational - http://www.justfm.org/about/ 62.56.89.38 11:40, 30 June 2008 (UTC)[reply]

Wikispecies[edit]

There isn't much point in having a "today's animal". The whole project is taxonomic; of little interest to almost anyone but taxonomists. No-one wants to here a read-out of a species' genus, order, family etc. etc.. ----Anonymous DissidentTalk 11:17, 31 May 2008 (UTC)[reply]

Launch?[edit]

When is the radio station going to launch?

There is no plan, the project has died. :-( Cbrown1023 talk 22:12, 5 March 2009 (UTC)[reply]
Make it live! --ShakataGaNai ^_^ 06:31, 3 September 2009 (UTC)[reply]

Back from the dead![edit]

I think it's time to revive this. I'm now on Ubuntu and I can safely say there's a lot more options than already on the talk page. For a test, it can be run off a pretty basic PC running Icecast. Yes, there are a variety of tools to automate playlisting and fallback to pre-scripted content. There's also the option to do live stuff. WikiVoices (which I'm in with Jimmy next week) was what reminded me about this. It would be perfectly possible to do the show live, record for editor cleanup, and then rebroadcast the recording when appropriate.

If running on Linux there's IDJC (Internet DJ Console). I've successfully run a two-hour broadcast with that streaming via a server on http://1Radio.org. IIRC their setup supports 100 users at 64kbps/user stream. So, ~6Mbps per 100 users. For controlling, I can do it from a 5-year old PC with only 1Gb RAM. I believe an XP setup of a similar age could, if spyware and Norton-free, would also manage. (This was PC->USB 1.1 WiFi Dongle -> Hub -> ADSL -> {Internet} -> 1Radio.org -> listeners.)

I'm asking around to see about getting a test server to support maybe 20-30 listeners for a pilot. That would, initially, be buggy and not always running. You'd likely get periods tuning into strange voices reading Wikipedia articles in a monotone; basically, a proof-of-concept. A key thing for that would be setting up an on-connect intro. Pretty sure Icecast Can't do that. I'd run it here but I share the DSL connection with others. I need something that could sustain 500kbps output reliably, and it needs to be accessed by people who are trusted not to broadcast something non-freely licensed (like top-20 hits).

If there's a basic proof-of-concept running on a loop then additional content can be tried, such as recording the WikiVoices sessions through it. I believe Asterisk can be tied into it; so you could do call-in/talkshow style; thing there is can you have a bleep/abuser delay? --Brian McNeil / talk 07:59, 17 December 2009 (UTC)[reply]

Questions for automation/test run[edit]

Mostly this is getting my hands, and some other people's on a running server.

  1. How can you request a stream of an audio file from Commons?
  2. Are audio files on Commons mandated to have accurate metadata tags?
  3. Can we work from "lists of lists", i.e. schedule 1 hour content to be repeated and slot in master schedule several times.
  4. Can these files be relatively easily set up?
  5. Can we set a "live broadcast" slot? ½-hour segments, compieré connects in 3-4 minutes before slot goes live (with password), broadcasts for ½-hour, back to schedule. If no connection 3 min. after slot start, have fallback programming.

Any more? --Brian McNeil / talk 08:08, 17 December 2009 (UTC)[reply]

Content suggestion[edit]

Erm Brianmc, How quickly could you make a copyright free playlist for Burns Night? XD (ShakespeareFan00 using alternate account) Sfan00 IMG 22:52, 26 December 2009 (UTC)[reply]

Possible code (Re)use[edit]

Firstly my apologies for my relative inexperience with editing Wikis.

Around 2010 I created a project called "TeamPlayer". Basically it was spawned from an idea that a colleague had. TeamPlayer is what I describe as a "democratic Internet radio station". Specifically the "about" says:

"TeamPlayer is a democratic Internet radio station. Members create song lists and the virtual DJ, DJ Ango, goes from member to member playing a song out of their list. Members get to hear what their fellow members' musical interests are, and contribute songs on their own. With TeamPlayer everyone's musical tastes are heard and everyone gets an equal chance to be heard."

We had used it at my work place for a while, running on a virtual machine, until interest (both mine and my colleagues) died out and the VM host went away. I left it at that and moved on. But a few weeks ago my creative juices started flowing again. I added some features to what I call TeamPlayer 2.0. Some of those features include the ability to have multiple radio stations, a "library" where songs are stored and can be re-queued later (by any user) without having to upload the file again, search for the library, notifications, podcasts (for the library/stations), etc. I limited the number of stations that can be created to a max of 1 per user. This was to disuade everyone creating multiple stations and putting a bunch of content on them but no one was listening.

Anyway I had been thinking of changing the name because "TeamPlayer" kind of implies that there is a ("team") leader and "player" implies that you have more control of the content than you actually do (there no skip, fast-forward, rewind, etc). So I wanted to emphasize that it's both headless and it's "radio" and not a music player. So that's when I thought of "WikiRadio". When I did an Internet search for "WikiRadio" I came across this link and started reading the ideas and, while there is not a complete 1:1 match, it seems that what is in TeamPlayer fits a lot with what is wanted for "WikiRadio". TeamPlayer is mainly a Django-based app and uses MPD to stream via HTTP. MPD can also talk to IceCast I think but so far straight MPD/HTTP has worked for me. It has a (mostly) single-page app frontend, but users can also listen to the http streams directly (without requiring a TeamPlayer account). Also it can scrobble to last.fm.

TeamPlayer was originally designed for music streaming. But I think it could be adapted pretty easily for other content types. I have not open sourced it (yet) mostly because 1.x used a lot of copyrighted images and such and I don't know the legality behind distributing code that can be used for streaming music. There's a few other reasons as well.

Anyway my basic question is does this sound like something Wikimedia might be interested in. --Slod Ion (talk) 15:25, 8 July 2013 (UTC)[reply]

A fine idea. Worth revisiting with modern tools.[edit]

In particular, having a radio user for music on the projects is incentive to add your favorite collections :) And now we have a much roader range of CC music on various platforms to draw from. –SJ talk  02:18, 26 May 2022 (UTC)[reply]

Retrieved from "https://meta.wikimedia.org/w/index.php?title=Talk:Wikimedia_Radio&oldid=23330034"