It’s inevitable that having mastered their own specialty, an artist will go on to explore other media. For example, a painter who has worked with oils will probably try watercolors. An experienced electrical engineer will probably think he’s capable of doing mechanical engineering. For myself, I’ve spent the last few years writing this blog and always wondered how it would be if I could do a podcast.
Given how much I hate the sound of my voice, this wasn’t going to happen. I could hire an artist to do my podcast but when you don’t make any money to start with it really doesn’t make sense to spend more money. I waste a few microseconds every year considering it but another dream shot down. Amazon Polly for WordPress might just change that.
This weekend a number of things came together. I was recovering from a cold, the weather here in Texas was wet and dreary and I found myself with some free time. While I was catching up on my Google feed, I saw an announcement that Amazon was making it easy to do podcasts for Alexa. Even better, the voice was provided by Amazon’s text to speech program, Polly.
I really like Alexa. My rural Internet situation means that I have to carefully consider the impact of every device I bring into my house. It says a lot that I actually have two Alexa minis in my house, although recent events may force me to rethink that situation. After suffering from my cold, my wife discovered that she could use Alexa to summon me.
All of which is to say, I decided to try the Amazon app and create a companion Opinionbypen podcast. Sometimes it’s not the destination it’s a journey you make get there.
Amazon has tried to make this has absolutely easy as possible. If you read their instructions, all you have to do is set up an AWS account, install the Amazon app on your WordPress site, and link it with Alexa. How hard could that be?
Setting up the AWS account was easy. Still, I won’t even pretend I understood the security permissions they had me set up. Signing up gave me five million character translations per month free for the first year. They also provide translation and storage functions but I decided to forego those parts of the journey. The WordPress app was free, and I was ready to go. If only it was that easy.
I wasn’t about to risk my live site so I started out with a local copy of Opinionbypen using a program called Instant WordPress. I’ve been using it for years to try out new features. Unfortunately, the version that I was using was too old to support the Amazon app and I had to download a new version. The new version was a radical change from previous version that I had been using and I found myself having to build a new website.
I was now ready to try out the Amazon app. As I said they make it as easy as possible. When you create your post, you check the little box at the bottom and select the voice. Saving the post automatically translates it into speech and creates an MP3 file. Well, almost but not quite. In my case, it would never translate my post. I could do a bulk translate which means you translate everything on your website and fortunately since I was using my new website, I had only two posts but I could not get it to create the speech file for an individual post.
I’m going to jump ahead the two days it took me to figure this out. I found a discussion between the developer of the app and another blog writer where the writer was asking for a button to cause translation on demand rather than automatically doing it when you save the post.
The developer did not quite understand why this was necessary and was more interested in helping troubleshoot why the app was not creating the speech file. Bottom line, if you have a slow website your results may be mixed. Because I was using instant WordPress the response time between the Amazon server in my setup took much longer than the app allowed before timing out. Moving the app to my live site allowed it to work as intended.
This is one of those times that I find history repeating itself. On my 3-D printer I use a slicer called Cura. When it originally came out, it would automatically slice your model every time you made a change. Except that under certain circumstances it would not slice. I never figured out why but Cura finally gave in and gave us a button to initiate slicing. I hope Amazon will ultimately find the same solution.
Back to my issue, I was able to do a bulk translate and was absolutely amazed by the quality of the voice. When I tried it using a British male voice my wife found my blog far more fascinating. So much for my Texas drawl.
The text to speech wasn’t without flaws. When the app encounters pictures, it uses the alt text description of the picture. I don’t use alt text on my pictures and this method only works if you have a large picture in line with the text. On Opinionbypen, I seldom put in a large picture. Rather, I use a small picture that I put to the right side of the text. The picture is more for illustrating a concept and an attempt not to break your reading flow. Because of the way WordPress is organized, the picture is actually positioned before the text on the left.
That means that the vocal translation describes the picture before the text associated with the picture. If there is no alt text description, the voice gives the name of the picture. Amazon gives you a means to ignore specific sections of text by inserting a tag before and after the text. It’s a necessary feature but they didn’t count on the issues that the tags cause, especially when used with a picture.
After inserting the tags, I found my post full of ugly white space and my pictures were now isolated between paragraphs. I could live with that but the idea of converting the entire Opinionbypen to a podcast would require a lot of reformatting each post.
Once again, when I went to the Polly forums, the developer was responsive but simply did not understand why anyone would want to skip the pictures. I bow to his skills as a programmer but he doesn’t have a background in WordPress blogging. Fortunately, he seems very responsive to user suggestions.
Meanwhile, because of the picture issue and the lack of a translate on demand button, I find myself with a difficult decision. Postpone the journey until the developer fixes the app, write all future posts without pictures or hack my blog by FTP’ing a replacement MP3 file created without pictures. No need to guess, you already know which option I chose.
Let’s drop to the bottom line. Before publishing, I typically make about four revisions to each post. Although free for the first year, converting this post would normally cost me about $0.15. Since I chose to host the sound files on my website, there is no reoccurring cost for traffic. Considering that I paid $70 for a text to speech program that doesn’t sound nearly as good, I consider this money well spent.
So how easy is it to connect to Alexa? Sorry, that’s the other limitation. A podcast is limited to ten minutes and I’ve used that up getting you here. You’ll have to check back next week for the competition of my journey.
PS: Those of you receiving this post by subscription will have to go to Opinionbypen.com in order to listen. At least until I get the Podcast details worked out.
© 2019, Byron Seastrunk. All rights reserved.