The other day I was watching “Alex & Emma“, a movie about an author who hires a stenographer to help him write a book in 30 days so that he can pay off his gambling debts and avoid bodily harm. As I’ve been writing more, I began to wonder what similar tactics I might use to increase my productivity. In the past I’ve used Android’s Speech to Text feature to quickly compose texts and I began to wonder whether I could exploit this feature to compose longer blog posts, at what rate, and at what accuracy.
As I consider myself a reasonable typist, I decided to perform an experiment to determine whether Android’s Speech to Text feature would be a viable alternative to typing.
I would take an online typing test to measure my typing speed and accuracy. The online test calculates the words per minute (WPM), characters per minute (CPM), and one can determine the accuracy based on the number of words typed and mistakes made.
I would read a page from The 4-Hour Workweek, one of my favorite books, aloud and have Android’s Speech to Text feature transcribe it. I would record the time using a stopwatch and derive the WPM, CPM, and accuracy from the results.
I would repeat both tests 3 times and average the results for comparison.
Below are the results of my online typing test. The average of the three trials showed 63 WPM, 344 CPM, and a 91% accuracy. According to the web site my average CPM is faster than 88% of the 21 million people that have taken the online typing test. So, I suppose I’m pretty fast.
Typing Test Results
Note: It’s unclear whether the CPM from the typing test includes or excludes spaces.
Below are the results from my reading The 4-Hour Workweek and having Android’s Speech to Text feature transcribe the information. The average of the three trials showed 166 WPM, 798 CPM, and a 91% accuracy! This is more than double my typing speed and the Speech to Text feature achieved the same accuracy. The CPM is literally off the chart and is faster than 100% of the 21 million people that have taken the online typing test.
Speech to Text Results
|WPM||CPM (Spaces Excluded)||CPM (Spaces Included)||Accuracy|
Note: Punctuation and returns have been excluded from the accuracy numbers such that the online typing test and the Speech to Text comparison would be valid. There are voice commands in Android’s Speech to Text feature, but I didn’t have time to get proficient with them. Plus there are no punctuation or returns in the online typing test.
Using Android’s Speech to Text feature to generate content quickly is a viable alternative to typing. In fact, even a non-typist should be able to generate high WPM and CPM rates using Speech to Text transcription. If you are a decent typist, then you can use Android’s Speech to Text feature to give your hands a break. Heck, you could even do it laying down and give your back a break.
Knowing what to say and how to say it greatly enhances the speed at which content is created. The rates achieved during this experiment were while typing or reading existing materials. For a descent typist, it’s likely that the productivity gain due to the faster Speech to Text transcription rate is marginal compared to the time spent thinking about an idea, researching it, and constructing the information. However, there’s definitely a novelty in composing the written word through transcription.
Based on the test results, I’m able to screw up 9% of what ends up in the first draft, whether I’ve typed the information or had it transcribed. It’s highly unlikely that the Speech to Text transcription or anything typed will be publishable without further revision.
I’ve been considering using the speech recognition software at work, but I suspect that may be too distracting… especially if everyone started doing it.
Android punctuation and line spacing voice commands are as follows:
- Period (.)
- comma (,)
- question mark (?)
- exclamation point (!)
- new line
- new paragraph