Talk:Strategy/Wikimedia movement/2018-20/Working Groups/Diversity/Recommendations/10
Normally, an interviewer must use CC-BY or CC-BY-SA to release an interview as "open content". However, an interviewee would be recorded and would not know much about copyrights and the licensing. Must an interviewee be notified about the licensing and copyright? Would an interviewee agree to allow an interview to be used commercially and agree to have it released under an "open content" license, like CC-BY-SA? George Ho (talk) 04:59, 11 August 2019 (UTC)
yes, other projects, such as Transcribe Smithsonian are beginning to transcribe their oral history archives, which are fine primary source material, and the finding aid should be a good citation. Slowking4 (talk) 19:39, 12 August 2019 (UTC)
The "read Wikipedia articles" part of this already exists and has existed since 2005, as w:en:Wikipedia:Spoken articles. What has happened with that project is that many important articles have been saddled with outdated recordings from 2005 to 2008, and more than 99.9% of articles will never receive a spoken version at the current rate. There are several reasons I think the project (at least on the English Wikipedia) has barely gotten off the ground:
- It takes time to record and re-record audio files. While personal computing devices have improved greatly since the early 2000s, it can still take quite a long time to read an entire Wikipedia article out loud. Given that an article may change every day, it would be necessary to continually re-record parts of articles or entire articles just to maintain the usefulness of the recordings.
- There are/were not enough volunteers. Most active contributors already have their hands full writing articles. While it's possible that this could be remedied by finding an alternate user base, serious work would need to be done recruiting enough volunteers to make this usable (i.e. the project would be in a state where all important Wikipedia articles have up-to-date recordings), and that energy could perhaps be better spent recruiting people to produce and improve content. (While this may also be exacerbated by the project being virtually unknown, it's not a given that it would suddenly become much more active if more people knew about it.)
- Because the project appears to be non-viable due to inactivity, very few new recordings are created (approximately 25–50 per year) because there is little incentive for volunteers to contribute to an abandoned project that will never be good enough to be actually useful to the average visually impaired person who wants to listen to an arbitrary article. This feedback loop ensures that the project will never be viable unless there is a significant investment into creating and maintaining new recordings.
- The only point in time that the project became worthy of mention by third parties was when a contributor uploaded a low-quality recording to w:en:Bhutanese passport; the recording went viral because the contributor spoke using a (possibly fake) Bhutanese accent and sounded as though their voice had been digitally manipulated. (The recording has since been removed from Commons and replaced due to its low quality.) While this is probably the only time this has happened, there are fewer than 2,000 spoken articles, so it's very possible that this sort of thing would happen quite often with the proposed project, and the project would need to maintain (ultimately subjective) standards for how good a recording needs to be to be retained on the site and shown to readers/listeners.
Articles can change and improve quickly, and it would take a lot of coordination and a lot of people with decent microphones to get real humans to record and re-record these articles. I think a more feasible way to approach this would be to develop either improved screen reader software or improved text-to-speech software, since this would take much less human effort and would be able to cover a much larger base of articles (albeit possibly in much fewer languages). I don't see much of a point in asking hundreds of volunteers to spend hours doing what would be menial tasks when existing software could reduce the time it takes to do such tasks by orders of magnitude.
A database of voice recordings (not being based on an existing Wikimedia project) could be significantly more useful, due to the inherent linguistic and ethnographic value of storing such recordings. However, if users were to focus on recording Wikipedia articles or Wikidata items, it would not really be a very useful ethnographic record, since the language used in the Wikimedia projects is usually very dry and factual, and so the project would fail to capture certain linguistic nuances (e.g. casual conversation, slang). A similar problem would arise if interviewers were to focus solely on interviews. Furthermore, depending on the scale of the project, being able to automatically transcribe and translate most recordings would be a necessity due to the possibility of numerous copyright violations being uploaded (e.g. from reading a copyrighted work or singing a copyrighted song), particularly since most people don't know more than two or three languages. Jc86035 (talk) 15:25, 13 August 2019 (UTC)
- Very good points. My impression of the "recommendations" was that they start to address a number of different issues, and then suddenly there is a proposal for a new Wikimedia wiki (why not simply present it on Meta in the normal process?). This all is still very unready, like the result of a brain storm session, but not a substantial proposal. Ziko (talk) 09:46, 14 August 2019 (UTC)
More ambitious: An improved Wikimedia Commons APP
I love the idea of making audio upload easier, because I always try to improve audio content. But in the age of Netflix and HBO We should be more ambitious: a Wikimedia Commons APP. Alice in Wonderland? Popeye the sailor? Then do it. Make it TV compatible and, well, maybe Wikimedia Commons has not the best film collection, but let's make files easy for users.--TaronjaSatsuma (talk) 16:31, 13 August 2019 (UTC)
What about sign languages?
Sign languages are not written, but they are not oral either. A project that is not based on written languages is the best opportunity they have to be included in the Wikimedia movement, since VideoWiki does not seem to be designed for that. Has the Working Group considered this possibility? --Unapersona (talk) 10:44, 14 August 2019 (UTC)