Open mattodd opened 4 years ago
Hi all,
Great idea. Youtube has good captioning software and so does the studio facility on Canvas and other LMS. It might be possible to use this and then export the videos or make the page public. Happy to help with this if anyone wants to discuss - we're working on some new videos for Breaking Good at the moment so can share some of the experiences with this too and will be putting out a call for some contributions too :) Hope everyone is very well, warmest wishes Alice
On Tue, Apr 21, 2020 at 8:25 AM Mat Todd notifications@github.com wrote:
A perennial problem in open source drug discovery projects: how to maintain an easily-understandable Story so Far - a quick, up-to-date summary of where we are and what's currently most important for people to look at.
It takes effort. The wiki for each project (repository) is the place for the full Story, but it can be difficult for people new to the project to glean the most important current threads. Project Twitter feeds can alert people to requirements. Well-written Issues can really help to solve things but can also take a lot of time to do well. All these aspects are democratized, in that anyone can do them, which is crucial. But writing well takes time, and, speaking personally, the activation barrier has often prevented me from posting something, even if it's clearly needed.
We need to experiment with something that helps explain needs/developments but which is complementary to what we're doing, and which:
a) is quick to do b) is quicker than writing c) can draw attention to the most important things that are needed, or can act as a summary of where we are with a part of a project d) can be easily digested by people in 30 seconds e) might bring new people in to help with key problems f) is interesting
Spend any time on public transport and you'll see the most common form of information dissemination is ... the short, captioned video (captioning means you can have the sound off). I think we need to try this to see if it helps.
We need a technical solution that allows anyone to create a 30-second video in which they are speaking, and which is captioned (the words spoken appear on screen) and the video can include a URL at the end for more info. Ideally the video could include an image or screen that highlights what the speaker is talking about. These videos need to be recordable and uploadable by anyone.
Some technical things needing an answer:
- How best to record a video? Zoom with screenshare? A phone with the images added after?
- How should we upload recordings? To one account (requiring shared access), or just with the right keywords and a logo?
- How can we automate captioning so nobody has to type like they're Hansard.
Can I ask @danaklug https://github.com/danaklug , @edwintse https://github.com/edwintse and @fantasy121 https://github.com/fantasy121 for their initial thoughts and gut feelings, and to ping others likely to be interested? I spoke about this with @alintheopen https://github.com/alintheopen about 2 years ago and discussed captioning. I found this https://www.washington.edu/accessibility/videos/youtube/ and rev.com, but that's as far as I got.
We'll need volunteers to try a few potential solutions.
Videos like this are no substitute for a solid wiki that accurately describes the detail of where we're up to, but they can highlight important developments quickly, can help people understand where we're up to, and can help emphasise better the very human element of much of the work that we do.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/OpenSourceMalaria/TechOps/issues/3, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAUBIJ3BUXVAVLPKE7LLAZ3RNTDWXANCNFSM4MM2ECAA .
I think it's a nice idea which would probably work best on Facebook, Twitter, LinkedIn or Instagram (all four?). There are lots of options for automatic captioning with different capabilities and price points.
YouTube, Facebook, Twitter and LinkedIn will all add captions to video you upload to those platforms. This may be easiest although as far as I can tell none of them will export the captioned video as one file, so it may be a bit more work to post something on multiple platforms and I suspect they would all look slightly different (not sure how much we care about this). Instagram doesn't have this feature as far as I can tell. The other option would be to post a link to the captioned YouTube video but I think it would be better if it were embedded because we could take advantage of autoplay and remove the added step of clicking on a link.
There are lots of apps that will caption videos but I think it would be difficult to get everyone using the same one as they are usually either/or with Android and iOS devices. There are also a variety of subscription services which I suppose we could potentially share the login to, or have one person in charge of captioning. A quick summary below.
My initial thoughts about posting these are that it's better to just have one or a few people in charge of the social media accounts (which I think is how they are run now?) It should be simple to email video files or even leave them in GitHub repositories and have them posted to the account from there. My only reservation is that this may turn into a lot of time for whoever is running the account that they'd rather spend on other things... but that's more a question of how we run social media in general.
Hope that helps!
We have made some progress on video updates and the first one is live on YouTube. The workflow so far has been fairly straightforward: the video was recorded on Zoom with screen sharing, then uploaded to YouTube. Captions were generated by using the auto caption feature and then manually edited.
The question now is how best to share the video. I've had a go at uploading it directly to Facebook, but ran into issues with the file type for the captions that is available for download from YouTube, and it seems like you have to pay to be able to have Facebook auto caption your videos. Any suggestions as to how to troubleshoot this or other ways to share would be great! Ideally we would like to have it auto play as someone is scrolling through their feed.
The video file (.mp4) and caption file (.sbv) have been uploaded to the OSM TechOps repository.
It's really good @danaklug @mbhebhe . When we write this up as a short protocol, let's emphasise the need for 30 seconds, 3 slides and the hashtags #OpenSourceMalaria #openscience #opensourcedrugdiscovery. That way there's some consistency and the videos can easily be found and grouped. Now if we can just solve the problem of how to post elsewhere. One potential solution already: https://twitter.com/le_rako/status/1280855719131009030
Hi all, Great video! Apart from Instagram, maybe the best way is to just stick to YouTube and to make sure that the link embeds in a way that the whole video shows up, rather than just a link. This way it is also easier to track metrics for the video from one source. Loving your work @mbhebhe and @danaklug :)
@alintheopen Thanks! :) I can get the video itself to show up on posts, but it's still linked and not embedded in whatever platform we're using to share. The only downside to this is that to watch the video people will have to click on it (it won't autoplay when someone is scrolling their feed). On the flip side, it's a good point about just having one source and tracking metrics from there.
@mattodd That tip seems to be for turning captions on by default when someone is playing the video through Youtube, which is definitely useful if we decide to just use the link but won't really help if we want make a separate post on another platform.
As a proof of concept I worked up some code to run the video from Tha through GCP's Video Intelligence API. The results would still need some editing - but it's free (for the first 1000 minutes / month) which would effectively give us a quota of 1000 vides (each minute is rounded up).
Converting it from the raw response to another format (e.g., SRT/SBV file) is reasonably simple.
One advantage of this is that it would give us the choice between closed subtitles (ability to switch on / off) and hard subtitles (always on).
Results below from my test run on GCP.
Transcript: Hi, my name is Tom and his part of Open Source. Malaria. I'm working on new medicines to combat this terrible disease. I recently got some new data back on the molecule. I'm working on that seems to suggest it kills and malaria parasite targeting an enzyme called Carbonic anhydrase. This is cool. This enzyme is a pre validated malaria Target and there are no drugs with this mechanism of action. We are now in the process of checking these results and digging further. Confidence: 0.9127715229988098
Transcript: As an office or science project he can participate in the discussion the chemistry and the biology and join in our search for new cures. Check it out. Confidence: 0.8296579718589783
A perennial problem in open source drug discovery projects: how to maintain an easily-understandable Story so Far - a quick, up-to-date summary of where we are and what's currently most important for people to look at.
It takes effort. The wiki for each project (repository) is the place for the full Story, but it can be difficult for people new to the project to glean the most important current threads. Project Twitter feeds can alert people to requirements. Well-written Issues can really help to solve things but can also take a lot of time to do well. All these aspects are democratized, in that anyone can do them, which is crucial. But writing well takes time, and, speaking personally, the activation barrier has often prevented me from posting something, even if it's clearly needed.
We need to experiment with something that helps explain needs/developments but which is complementary to what we're doing, and which:
a) is quick to do b) is quicker than writing c) can draw attention to the most important things that are needed, or can act as a summary of where we are with a part of a project d) can be easily digested by people in 30 seconds e) might bring new people in to help with key problems f) is interesting
Spend any time on public transport and you'll see the most common form of information dissemination is ... the short, captioned video (captioning means you can have the sound off). I think we need to try this to see if it helps.
We need a technical solution that allows anyone to create a 30-second video in which they are speaking, and which is captioned (the words spoken appear on screen) and the video can include a URL at the end for more info. Ideally the video could include an image or screen that highlights what the speaker is talking about. These videos need to be recordable and uploadable by anyone.
Some technical things needing an answer:
1) How best to record a video? Zoom with screenshare? A phone with the images added after? 2) How should we upload recordings? To one account (requiring shared access), or just with the right keywords and a logo? 3) How can we automate captioning so nobody has to type like they're Hansard.
Can I ask @danaklug , @edwintse and @fantasy121 for their initial thoughts and gut feelings, and to ping others likely to be interested? I spoke about this with @alintheopen about 2 years ago and discussed captioning. I found this and rev.com, but that's as far as I got.
We'll need volunteers to try a few potential solutions.
Videos like this are no substitute for a solid wiki that accurately describes the detail of where we're up to, but they can highlight important developments quickly, can help people understand where we're up to, and can help emphasise better the very human element of much of the work that we do.