# Opencast Summit 2022 – Session Abstracts A list of all submitted sessions for the 2022 Opencast Summit. Visit [opencast.org to get more details about the summit](https://opencast.org/2022/02/25/opencast-summit-2022/). ## A subtitle editor for the editor > Presenter(s): Arne Wilken, Daniel Ebbert Opencast allows uploading and displaying subtitles for videos, and can even generate subtitles during processing. However, there is no easy way to edit subtitles once they are uploaded or created. Educast NRW and ELAN are aiming to change that by adding a lightweight subtitle editor to the new video editor. Work on the subtitle editor has just begun, so this talk will focus on what our goals are with the subtitle editor and what the user interface might look like. ## Analysing videos using the Annotation Tool: New features and next steps > Presenter(s): David Graf The latest developments of the annotation tool bring new didactical possibilities and simplify the use of the tool. In this lightning talk the following new features will be presented: 1. New sharing option for annotation tracks: «Visible by default (automatically shown to all users)» 2. New option «Pause video: When setting a structured annotation» 3. Implementation of modals to create/edit categories and scales If you want to learn more about the basic functionalities of the tool and its use cases please see the following video: https://video.ethz.ch/events/opencast/2018/vienna/45db0d29-64bf-4984-bf56-9379fb06b44b.html ## Beyond Elasticsearch: user-facing search engines > Presenter(s): Lukas Kalbertodt Quickly finding what you are looking for is oftentimes a vital part of a website's or app's user experience. However, building a good search feature is everything but trivial: the `LIKE` filter in SQL doesn't get you very far. This is where search engines come in. They allow you to quickly search through vast amounts of data and filter, sort and rank the results. Different search engines focus on different use cases and thus, some are better suited for user-facing search than others. While for many, ElasticSearch is the standard solution, it might not be the best choice in terms of user *and* developer experience. ## Capture Agent Operations > Presenter(s): Paul Pettit We are planning a complete overhaul of how we manage our 350+ capture agents. It would be great to discuss ideas, experiences and best practice around deployment, management, monitoring etc. including hardware, OS deployment/updates, remote support and anything else you can think of. ## Combining AI & Humans to allow video-accessibility at scale - hosted by Amberscript > Presenter(s): Carolina Kluck ## Easier hybrid teaching > Presenter(s): Rüdiger Rolf Opencast makes regular lecture recordings easier. But after the pandemic we see an increased demand hybrid courses. That means that some of the students are in the classroom, while other participate remotely. Additional to simply recording or live-streaming the course for the remote participants, lecturers expect more interactivity, like in videoconferences. But using regular videoconferencing tools makes the job of a lecturer quite stressful, as there is much to manage during the lecture and it will get hard for the teacher to focus on the content of the course. Within this talk we want to present currently running developments from the Osnabrück University to improve BigBlueButton and Opencast for hybrid teaching. ## High Available Opencast > Presenter(s): Lars Kiesow, Alex Laschinger Talking about high availability with Opencast, pinpoints and further developments… ## Improving accessibility through automatic captioning (Hybrid and Single-Speaker models) > Presenter(s): Graham Morrissey Way With Words currently provides UCT with an automated captioning service, which is accessible via their LMS. The service is delivered in a hybrid fashion: 99%+ accurate, human-generated transcripts are force-aligned by our technology partner before going through QA and then delivery back to the client's LMS where students can access transcript and caption files along with their lecture content. In addition to the above, we've run a pilot project on a single-speaker model that can produce 95%+ accurate transcripts in an automated fashion after some initial transcription and training. We believe this to be a more scalable and affordable option for universities that might like to have speaker models for established academics that would like to have their lecture content automatically transcribed and captioned. We would appreciate the opportunity to present at Opencast Summit 2022. Please do let us know if any questions at all. ## Lecture2go: powered by Subtitle2go > Presenter(s): Benjamin Milde, Martin Kriszat We present an automatic subtitling solution for videos, that we integrated into Lecture2go, University Hamburg’s central media platform. Since 2008, it has been possible to watch, listen to and download university events recorded using video technology. Recently, we implemented automatic subtitle generation to make lectures more accessible. No external API calls or service providers are used for our speech recognition, as we train our own automatic speech recognition (ASR) models with the widely used open source toolkit Kaldi. This means we can offer the automatic transcriptions at no extra cost (other than computation) and can customise the ASR model and lexicon. We implemented additional algorithms for subtitles segmentation and trained additional models for punctuation reconstruction. The automatically generated subtitles could also serve as a basis to make the video material more accessible (e.g. via search, keyword clouds, and the like) and can be used for further manual revision, helping in significantly speeding up manual work. We implemented an integrated subtitle editor, that can be used to correct the automatic transcriptions. Our presentation gives a basic overview over the processing pipeline that use, how we created our own models and how they work in the context of Lecture2Go. Our Python based software might be a good starting point to implement and offer similar functionality in Opencast. ## Opencast Annotation Tool - A common way forward > Presenter(s): Daniel Ebbert In this session I would like to discuss a common way forward for all parties interested in using the Opencast Annotation Tool (OAT). Discussion points would be: - What is the current state of the OAT? - What needs to be done to get the OAT to a stable state usable by everyone? - Which feature request are of interest to multiple parties? - How can we coordinate future work on this tool? ## Opencast Ansible-Roles: Building Blocks for Your Setup > Presenter(s): Timo Nogueira Brockmeyer Opencast is a flexible configuration monster. But setting up Opencast from scratch can be pretty easy if you have the right building blocks. In this lightning talk, i will give you an overview over the Opencast Ansible-Roles the ELAN has published on ansible-galaxy and on how to use them to quickly build a multi-node setup. ## Opencast Architecture and Dependencies > Presenter(s): Greg Logan Organizer note: I am only reliably able to present on Wednesday, and would strongly prefer to be as late as possible. I have to get my kids up and out the door to daycare, which doesn't open until 1400 UTC. If this time doesn't work I can build a longer version of this talk as a workshop. In this talk we will be going over how Opencast's jobs work in detail. Expect to come out of this talk knowing how jobs are dispatched between nodes, why Opencast dispatches jobs the way it does, and how all our 3rd party dependencies fit into this. This talk will require some background knowledge of basic Opencast usage, but is intended to be understandable to a non-technical Opencast administrator. ## Opencast Archive on AWS S3: integrity and checksum fun! > Presenter(s): Paul Pettit After moving almost 2 million assets into AWS S3 as part of our cloud migration, we experienced a number of issues, mostly involving mapping of resources shared between mediapackages. In this talk we will explore some of the problems and solutions we found to restore and maintain the integrity of the archive. ## Opencast PHP Library and new LMS API > Presenter(s): Till, Farbod? ## Opencast in AWS (and other clouds) > Presenter(s): James Perrin Discussion about operating Opencast on AWS. Topics to discuss: deployment, operation, storage, delivery, new services. Though focus on AWS ,open to other cloud/local clouds and those considering migrating/deploying on a cloud provider. ## Ownership of Videos > Presenter(s): Lars Kiesow, Olaf Schulte Right now, Opencast has no concept for ownership of videos. You either have write access or you do not. Nevertheless, ownership is a concept that comes up from time to time when talking about upload of user content. We would like to start a discussion about this topic and talk about how this should work, what implication this concept has for things like access control and how we can make this happen in our code. ## Paella Player 7 > Presenter(s): Carlos Turro Just to present the features and changes in the new Paella release, as well as the new repo structure and the Low Latency HLS support. I would like to make it as interactive as possible, and it probably coulf fit in less than the 30 minutes slot ## Plugin-Management > Presenter(s): Lars Kiesow A new plugin management feature for Opencast is currently under development. The feature will allow users to easily enable or disable certain parts of Opencast. This is similar to (and in fact uses) the Karaf feature configuration, with the huge benefit that users do not have to configure exact versions and keep that updated. Instead, they just need to specify the plugin they want to load, and they can do that even at runtime. This talk will give a short introduction as to how this feature will work and what it can be used for. ## Push forward OC's accessibility > Presenter(s): Clemens Gruber We'd like to improve the accessibility of Opencast! In this BoF we'll try to collect recent and ongoing activities in the accessibility field and try to find supporters or brothers / sisters in mind that are interested in this also. Targets are the user facing components like player, portal, studio. What was done already? What activities are ongoing? What is a low hanging accessibility fruit in Opencast? What is important to tackle? Join the discussion and push OC forward for disabled and all people! ## Roll forward previous lectures > Presenter(s): Franck Tanoh Due to the Covid-19 pandemic, lecturers have had to manually upload dozens of videos instead of the usual automatic lecture capturing. As such, a good proportion of them wanted to re-use previously uploaded videos for the current year. This talk will highlight the in-house Opencast integrated tool we built to support them. ## State of the Crowdfunding > Presenter(s): Greg Logan, Lars Kiesow What’s the current state of the crowdfunding? Where are we, when it comes to our goals? What work has been done already? These are the question this session tries to answer. ## State of the new Admin UI > Presenter(s): Isabella ## The Great Manchester AWS Migration > Presenter(s): James Perrin University of Manchester runs one of the largest Opencast deployments with over 350 CA agents. Our Opencast deployment processes ~50,000 videos a year and maintains 250,0000 published videos, ~2.5TB of data. After more than 3 years of planning and delays last summer the whole Opencast backend and video portal was moved from on premise to AWS. This talk we summarise that process and what we have learned in our first 6months of operation. ## The challenges of supporting a lecture capture service > Presenter(s): Shazia Mohsin A talk about some of the challenges faced in supporting a lecture capture service from start to end and adapting to the shift from traditional face to face teaching, to a more flexible approach. ## Update on Tobira (new video portal) > Presenter(s): Lukas Kalbertodt Our regular update on the development of Tobira. Topics: new features, future development timeline, planned deployments and more. There will also be lots of time for questions and discussion! ## Virtual Capture Agent > Presenter(s): Ypatios Grigoriadis In order to enable automatic ingestion, assignment, and processing of quasi arbitrary video recording sources, a new feature has been developed: an enhancement of the OC-Inbox module which extracts metadata from incoming recordings and tries to match them to a scheduled event. ## Vosk > Presenter(s): Martin Wygas Opencast 11.3 introduced the generation of subtitles via Vosk. First, this talk will explain how to set the feature up and how the workflow operation handler can be configured. After that, we would like to know if somebody is already generating subtitles with Vosk and what the experiences are so far. ## What is the importance of open-source software ant universities? > Presenter(s): Rüdiger Rolf The reasons why universities use free and open-source software (OSS) might be very different: lower cost, more privacy, scalability, adaptiveness, and much more. And it seems that there are some trends regarding OSS. Institutions that already use free software probably tend to use it for new projects too. Others seem to be scared of OSS because of a lack of service. Or they don't understand that although it is free software, it does not come without any cost. If something is urgent one group is using open-source as they don't have to get a license first, while the others consider that is quicker and more reliable to buy a commercial solution. I would like to discuss with the community, what the status of OSS is at their institutions. And hopefully we will find some hints what this could mean for the future of Opencast. ## What's new in Opencast 11 > Presenter(s): Jonathan Neugebauer This talk will deal with the new features and enhancements in introduced in Opencast 11. ## Zooming into Opencast - 1 Year Later > Presenter(s): Greg Logan Organizer note: I am only reliably able to present on Wednesday, and would strongly prefer to be as late as possible. I have to get my kids up and out the door to daycare, which doesn't open until 1400 UTC. I could prepare this as a video if video submissions are allowed. A quick summary of the Zoom Ingest tool's status and features, along with how its production use has gone at ETH Zurich. ## educast.nrw: State of the project > Presenter(s): Daniel Ebbert In the beginning of 2021 the educast.nrw project started. The project is publicly funded and aims to provide Opencast as a service to all public higher education institutions in the state of North Rhine-Westphalia. The project is currently in a beta phase and some institutions already started using the service. This talk will provide an overview of what has been achieved so far and which contributions educast.nrw has, and is planning, to make to Opencast.