How to remember everything during online meeting

by Dmitry Kirsanov 25. October 2021 08:01

Pandemic brought one thing that changed the way we go through our daily office lives. And that’s – endless online meetings. At some companies you may be lucky to have just one per day, personally I’ve got the worst possible experience, as far as online meetings go (or, at least, my imagination refuses to show how it could be any worse) – non-stop 9 to 5.

Yes, such experience would be a 100% fault of your employer, and it’s not what you should be going through. A day of such experience draws all energy out of you, and often you may remember only as much as there was written in your calendar and notes. But there is a light at the end of this tunnel.

As a preface

Which can be skipped, as it explains how I got there.

When I’ve got a contract at (central financial regulator of the country) as a solution architect, I thought I’ll be doing exactly that. Creating software designs. Yeah… Turned out, they needed a secretary, to figure out what they do, and how projects interact with each other. Why a “secretary”? Because I had to sit through meetings of every team they had, for over 20 different projects. Then look into documentation of each project. Then I had to speak to different people from these teams, and produce trip reports about this experience.

After each day, I felt spent, wasted, there was no energy left. Additional problem was that mentioned organization is a complete mess in terms of structure, so if you think that your company is genius at creating problems out of thin air – think again, there are better examples. In other words, it was a road to nowhere, and I started to wonder – am I the only lucky man alive, or do others get through the same circles of hell?

Turned out, they do.

Two things came to my mind every day during this crazy experience. My experiments of 2014-2015 with neural networks, and what others said about their jobs as IT managers. I came to conclusion, that this part of life can be automated, and it definitely should.

So, after leaving that asylum and their crazy experiments, I joined my friend in what we called a Lab, to find a solution to the problems, brought to us by newfound digital life.

The problems of “digital life” as we saw them

One problem is to sit through these meetings, and get something out of them.

Second – to sit through your meetings, where you have to show slides and narrate them. It wasn’t such a problem for me, as I had no presentations to give, but I saw how other people were struggling through it. Onboarding, various trainings that go in circles – it had to be automated, somehow.

Another problem is way more complex, and stands on shoulders of two beforementioned giants – to understand what’s being said, dissect it and help you to answer questions.

Next gen problem – to ask questions for you. But we’ll get there too.

Solution to attention span problem during online meetings

Lies in transcoding it into text. All you have to do, is to create audio or video recording of it, using one of many tools available – from screen recording to a mobile phone next to your speakers. Then, you’ll put that record through speech-to-text solution, and there you go – a full transcript of your meeting, with no word lost about that wonderful trip your colleague had, and their plans for next weekend.

First, let me show you, what we made for you, to solve this problem. It’s a short video presentation, but it shows the whole process of converting your recording to speech. Doesn’t take long, and is completely free. Then, I’ll explain how it works, if you are interested.

So, here comes the explanation.

There are two ways to transcribe your recorded speech into text – offline and online. I am currently working on bringing the offline way to general users, but that will require some real dedication from user – over a gigabyte of hard drive space and powerful enough processor. Not everyone has it, not everyone can install it, as sometimes it’s done in somewhat restricted environments. But it can be done. There are few engines for it, including free ones, and one of them is Mozilla DeepSpeech. To make long story short – every second word will be transcribed wrong, and if you speak in other language, your chances are dim. It will improve, and we’ll do everything in our power to help them and promote the project, but as of now, it’s not your favourite tool for the job.

There are also some paid cloud solutions. Basically, every cloud provider, plus some scientific clouds, in different countries, are providing these services. That’s cool and stuff, but only if you are serious about cognitive services. I’ll give you an example – I have registered API access for cognitive services in Microsoft Azure to run a few tests. In two weeks, I didn’t run a single one. I was billed EUR 300.

So, the choice number one for everyone, at this stage, should be a free tool that won’t result in a surprise bill, and that will do the job. And after numerous tests (and years in industry, yes) the winner is “good old” YouTube. They are using their paid API in own product for free, to add value to video hosting, and that’s what could be exploited. Or, if you don’t like the word – “leveraged”.

The only problem is – to use that tool is tricky, as we don’t want to actually upload our video recordings, wait for hours, recode our files and stuff. We need a one-button-solution. Hence comes the SoundWorks. It’s a free app that you are free to enjoy.

SoundWorks was made with completely different goal in mind. To create narrated presentations and use various ways to recognize speech. However, it includes all the tools that we need to “leverage” YouTube as our speech-to-text solution. And the video above is showing you exactly how to do it.

The video was narrated using SoundWorks, by the way, and the way it sounds in English is better than what Google or Microsoft could provide, and I tried. But we’ll speak about it soon.

Things to consider, when transcribing your videos

  • Security. Don’t let others download it (so, don’t publish your files), don’t leave video stream intact (replace it with still image, or decrease quality to unbearable minimum) and remember, that Google will have access to this text.
  • Local laws. Usually, it’s not a problem. If you can hear it – you can record it, and there are so many ways to do it – from phones and computers to smart glasses – that there is no way you can tell if you are recorded or not. In most countries it’s not an issue, if you are legitimate participant, but in some – it may have nuances. In other words – don’t get caught.
  • People. They usually don’t like knowing they are being recorded. That’s why we are working on stream transcriber within SoundWorks, that, technically, won’t record anything but text. However, that thing won’t be free, as it takes a lot of processing power and cloud resources, so if you want to do it for free, then recording is a necessary step.
    You can, however, apply some good will rules to it – don’t keep your records for long (you can store them in temporary locations, for example), don’t keep the video stream (screen recorders, like Camtasia or ShareX allow you to record audio only) and don’t boast that you record.

That’s about it, enjoy the time you have freed from making notes!

blog comments powered by Disqus