A Study in Content

For the last couple of weeks, I've been thinking a lot about my last blog post where I discussed the accessibility of podcasts. As discussed, the hearing impaired are a massive audience that, for the most part, are missing out on some of the greatest content on the internet. If we're going to make podcasts more accessible, providing a transcript of each episode just isn't good enough. We need to make the creation, distribution and consumption of these transcripts as simple and easy as it's audio counter part. With these three aspects in mind, let's start by taking a look at how podcasts are curated.

When it comes to hosting and publishing a podcast, services such as SoundCloud, Stitcher and Podbean seem to be at the head of the pack. These services have rich feature sets like a website builder, both iPhone and Android apps and analysis tools so you know how your podcast is performing. Podbean even has an advertising marketplace where you can connect with potential sponsors for you podcast and an app marketplace so third party developers can build plugins. While all these features add lots of value to each offering, the most important feature of all these platforms is RSS/iTunes support. Why is this such an essential feature? Well, I can't speak for everyone but I have my phone setup to automatically download the latest three episodes of my favorite podcasts. This is an incredibly convenient setup as I don't have to do anything in other than open my podcast app and subscribe to a podcast. This allows me to have a queue of audio content to listen to while I'm on my way to work or while I'm at the gym. The only way for me to do this is via an RSS feed which is used by my podcast app.

To accomplish our goal of reaching a wider audience, I believe we need to apply this workflow to the distribution of transcripts. Podbean's app marketplace seems to be a great place to get started. Using their API's we could automatically transcribe new podcast episodes. Once the transcription is done, we can than append the transcript to the episodes description. This would be a great start but falls short of fulfilling the workflow outlined above. In order to automatically distribute the transcript, we'll need to push it to a subscriber. This leaves us with a few questions that need answering, specifically:

1. Where do we push the transcript to?
2. What format do we push the transcript in?
3. How does someone subscribe to this feed?

I believe we can answer the first two questions with one word, eReaders. With eReaders such as the Kindle and it's associated tablet apps being so popular, I think it's only natural to push the transcript straight to a subscribers favorite eReader or tablet. As for the format, the Kindle and many other eReaders support the ePub format. Tablets, such as the iPad, also support the ePub format via iBooks app but in order to get the most reach, I think we'd also have to offer the transcript in PDF format. This leaves us with the question of subscription. Amazon has multiple ways of getting documents onto not only the Kindle but to pretty much any Android or iDevice via the Send to Kindle. In order to automatically push the transcript, we could ask a subscriber for their Kindle email address and automatically send the transcript once it's been produced. If the subscriber doesn't have a Kindle or doesn't want to use the Kindle app, we could automatically upload the transcript to a subscribers DropBox, Google Drive or OneDrive in either ePub or PDF format. In order to collect the subscribers Kindle email address or connect to any of these services, we'd provide a widget that a podcast producer can place on their site.

So in the words of The Black Eyed Peas, let's get it started!

A few days ago, a colleague of mine contacted me about the accessibility of podcasts for the hearing impaired. She knew that I had co-founded a service that automated the transcription of audio into text and thought it would be a great idea if all podcasts offered a text transcript of their episodes for the deaf. I'll be honest, while that is one of the benefits of having a text based transcript of your podcast, I never looked into what percentage of the population is deaf. So I did a little research and as per the World Federation of the Deaf, there are approximately 70 million hearing impaired people in the world. That's a rather large audience that's unfortunately missing out on the enjoyment of podcasts.

Being a web developer for many years, I knew that we haven't really done a great job of making rich content very accessible. I tried to find any complaints about the inaccessibility of podcasts to the hearing impaired. I came upon this article from 2015 on Time.com titled "Can We Build an Internet That Includes the Hearing Impaired?" While the article doesn't mention podcasts specifically, it does provide a very clear description about how poor the experience is trying to consume audio and video content on the web as a hearing impaired individual. Speaking from personal experience, the author, Steve Friess makes the argument that those who rely on text as a replacement for audio "are being left behind, stuck reading recaps, tweets, and live-blogs instead of experiencing key cultural and news events firsthand like everyone else." The author also makes the argument that "An awful lot of people are going to start missing out on an awful lot of stuff." I'd argue that an awful lot of people are already missing out on an awful lot of stuff.

With this in mind and considering most podcasts are episodic, the hardest part of producing a transcript of a podcast is turnaround time and accuracy. Manual transcription is usually very accurate but takes a long time and, depending on the length of the podcast, can be cost prohibitive. The Time article makes the argument that while automated transcription services exist they're not accurate "But YouTube also provides an automated captioning service of such poor quality that it creates more, not less, confusion." While that may have been true in 2015, services such as Microsoft's Cognitive Services and IBM's Watson have come a long way in terms of accuracy. These advancements can be seen in automated transcription services, such as Gumshoe, where we can deliver accurate transcripts very quickly. 

Last but certainly not least, the article asserts "Either way, someone needs to get on this sooner rather than later. Most hearing loss is irreversible and inconvenient..." While this is an unfortunate truth, we're working hard to help ease the production of the transcripts. We will also deliver a richer set of features that will enable anyone with a podcast make it more accessible.

We can and we will do better.