A friend recommended a book called “Captain Blood,” so I looked it up on wikipedia to find out more about it. On the wikipedia page there were links to free text and PDA versions at Project Gutenberg as well as a free audio version at Babblebooks.
I hadn’t heard of BabbleBooks before, but it sounded like a great idea. All the books aren’t free, but they have a few dozen free MP3 audio books. The audio books aren’t read by people, and they explain that this is a good thing because you don’t have to worry about too many theatrics or strange accents. An additional benefit is that you can speed them up to get through it faster without the voice sounding like a chipmunk.
I listened to the first few seconds of Captain Blood and all of their talk about how great it was to have text to speech software read the book instead of a person became meaningless. It was awful. It reminded me of my first experience with a Mac. I typed in humorous comments about a friend sitting next to me in the computer lab, then had the computer say it out loud. Hilarity ensued and he responded in kind. We found that you couldn’t spell words normally because they often sounded weird, so we had to change the spelling dramatically to get the right affect. For instance, to make the computer laugh I couldn’t just type hahahaha, I had to use haaaah, haaaah, haaaah or something along those lines.
In any case, you can be the judge of the quality, but I’m not impressed. Here is the Babblebooks version of Captain Blood chapter 1.
That reminded me of the free online audio that the LDS Church had made available on their site. I was impressed with how fast they made new articles available and the sheer amount they’ve made available. I showed it to a friend in college who was constantly listening to audio books and he claimed it wasn’t an actual person reading it. I listened to it again and couldn’t believe it wasn’t a human. It had the right voice inflections, there was a clear Utah accent and it was far better than any text to speech audio I had heard. I couldn’t believe that the technology had progressed so far that you could barely distinguish a human talking from a computer-generated recording. He continued to say he was sure it was automated and that he couldn’t believe I couldn’t tell. He explained that there are only a finite set of sounds in the English language and it’s just a simple matter of stringing them together in the right order to form words and sentences. A few others were with us and he convinced them of his opinion, and I finally gave in, saying it was the most impressive text to speech I had ever heard.
After reading Babblebook’s hype about how good their audio books were and then hearing how amateurish they sounded in reality, I decided to try out some text to speech software on my own to see how far the technology had advanced in over 10 years. I tried free versions of Natural Reader, Browse Aloud, Hal Reader and TextAloud and sure enough they all sounded just like that old Mac.
The way I see it, there are two explanations for the quality of the audio at LDS.org.
1. The LDS church is using incredibly impressive and advanced Utah-accented text to speech software.
2. The LDS church is using real people to read the material (who have Utah accents).
I just can’t see #1 being the case, so I am now firmly of the opinion that #2 is the reality. Additional evidence is that they already have everything they need to do high quality voice work, so it doesn’t require much effort because they have all the equipment they need to have someone read the latest articles and talks.
However, you don’t have to take my word for it. I selected the best-sounding software of the bunch (Hal Reader). I’m sure you will see that it’s laughable compared to the one from lds.org.
Since the automated ones are pretty bad still, I looked for some other sources of decent quality, human narrated, free online audio books. Here are the top 3 sites I found.