This is a “how it was done” post about our 4-minute long “Platform of Trust Developer eXperience” video. So far we have been missing a video introduction to Developer eXperience. It was not seen as a must-have before. Lately, the platform has gained the attention of organizations with hundreds of developers and thus now the introduction must be offered. Take a look at it and give feedback! Below the video, you’ll find the footsteps I took to get it done in cooperation with 3rd party developers.
As the Chief Development Officer (CDO) at Platform of Trust, one of my main goals is to increase development speed and part of that is to seek out new options in tools and processes. Initially, I was looking for efficient methods to get our written use cases, developer guides, and other material into video format. Generating voice-over with neural network-driven text-to-speech Amazon Polly was tested with various texts. The reasons to explore as few human resources as possible including model were:
- reduce development time
- more slick iteration possibilities
- get “native” sounding voice-over to the videos
- cut costs.
It felt natural to start from the Developer introduction video since now the time is right. I fiddled with Amazon Polly with several text snippets and eventually wrote the text with subtitles for “Platform of Trust Developer eXperience”. Then I created the voice-over with Polly using their neural network-driven solution. I did not use any SSML tags to fine-tune anything. After that I combined some graphics, the voice-over, and ready-made animations in Camtasia video editor. It took about an hour to get the initial introduction video done. (sample)
Now the initial video was created, but none of the real developers had seen it yet. Since I have quite an extensive developer network, I decided to test drive the video. I engaged in Twitter Viljami Kuosmanen (@anttiviljami) and Jani Karhunen (@janik6n). It's scary and thrilling to ask feedback from developers but it pays off. They give you brutally honest feedback and you will not make a fool of yourself. Feedback guides me in the right direction and helps me to make more likely the right decisions.
Their feedback was brutal but inspiring: “that steady synthetized narration makes concentration hard”, “speak it your self instead”, “as such the synthetized voice-over is a big thumbs down for me and raises the feeling of low-cost and hobby approach”, “video has good and clear content. Sound is really monotonic and to be honest I would not listen to it too long”
So we have the right content for developers, but voice-over sucks. Keep in mind that these developers are not our internal developers, but hardcore professionals mostly not working with Platform of Trust. After having discussions about how to improve the quality and still keep the costs low, Viljami suggested getting the voice-over from freelancers in Fiverr.
Using Fiverr was on my desktop in the earlier phase but I wanted to see if Polly would be able to produce good enough voice-over. Jani who has more experience with Polly said that yes it can create adequate quality but it still requires a lot of work (with SSML).
I raised the issue that we have around 30 videos coming and each needs to have voice-over. Some kind of hybrid development model has to be created since outsourcing videos totally would become too costly and reduce the agility.
In brief, we ended up in the development model in which Polly is used to test-driving the content in the video, which forces us to proofread and really fine-tune the content before proceeding. We can create Proof of Concept or even MVP-ish video without extra human resources by using Amazon Polly and Camtasia. However, high-quality output requires human effort. Since none of us in Platform of Trust is native American or British, we have to outsource voice-over to get quality results.
I was about to order a voice-over from Fiverr when Viljami said that we can use his voice and he’ll record the voice-over tonight if I provide the text. Wow! Even if this does not provide us viable option to create 30 videos, we get one of the most essential introduction videos with human voice-over.
The cherry on the cake is that since Viljami is a developer, the profound idea of “developer to developer” is injected in the video. A developer knows what to emphasize and draws attention to the most valuable parts. Viljami did some fine-tuning in the text as well and returned the voice-over after a couple of hours. After that, I just replaced the voice-over track in Camtasia and rescheduled some elements in the timeline. The result is just awesome!
Asking for feedback from developers who you would like to love your content pays off. Putting your neck in the line and get out there with your PoC is the way to go. Using your time in polishing something that has severe flaws is not going to make it any better. Push your creation out to the customers as fast as possible. That is scary, but they will tell you if it works and how to make it better.
I asked for feedback, but gained much more! Developers want to help. If they see value in your efforts, they are willing to put their time (or even voice) into it, even during Saturday evening as this case proved.
Any voice-over is not good enough if you are looking for high quality. Even if the voice-over is done by a native English speaking person, you might not gain the best possible result. Keep in mind that the last 5% differentiates great results from good results. Thus get the voice-over from a developer, not just from any native speaker. That brings the needed 5% final “developer to developer” touch and feeling in your video.