This just came across my desk as things do from time to time and it's something that a few of you might just find interesting as I did after reading it.
David Ferrin http://www.jaws-users.com *** Listening Speeds Approximation with Window-Eyes and JAWS Screen Reader Practice Text Documents Listening Speeds Approximation with Window-Eyes and JAWS Methodology In order to determine an approximate words-per-minute test for a screen reader, we used this process: 1. The screen reader's speed was set to a chosen percentage. 2. A long document was selected of sufficient length for the "Read All" command to be invoked without running out of text in a two-minute reading session. 3. A timer was set for two minutes. 4. The timer and the "Read All" command were invoked at the same time. 5. The "Read All" command was stopped when the timer expired. 6. The rest of the document was selected and deleted. 7. The Microsoft Word "Word Count" command was invoked. 8. The number of characters without spaces were found. 9. This number was placed into Windows Calculator. 10. The number was divided by 5 to obtain an approximate average length of five characters. (Five characters is the average word length in American English.) 11. The result of this was divided by two to obtain average words per minute. 12. The deleted text was restored and the process repeated for another speed rate. Known factors which influence speed outcomes: 1. By default, the screen reader pauses for a fraction of a second at sentence punctuation marks, such as commas and periods, in order to give a better reading experience. This would tend to give a slightly slower average speed for each percentage. 2. In calculating the number of characters, punctuation of all kinds, including quotation marks, dashes, and sentence punctuation and other printable characters were included (spaces were excluded). This would tend to give a higher character count, and therefore, a higher average speed. Though this has not been tested, the assumption is that the two variables tend to cancel out one another. 3. Different versions of the Eloquence synthesizer have yielded different results. Both Window-Eyes and JAWS have included a proprietary version of Eloquence within their screen reader in recent releases. These versions have yielded higher rates from the nonproprietary versions. 4. Probably related to the previous point, we found that current versions of these two screen readers produced markedly different reading rates at the same percentages. The test was performed for Window-Eyes 7 and JAWS 11 with their own proprietary version of Eloquence. Comparison Tables Percent Window-Eyes 7 JAWS 11 60% 346.3 524.7 50% 251.6 437.4 40% 196.5 344.5 30% 135.3 284.1 20% 107.8 224 Percent NVDA 95% 400 58% 300 27% 200 NVDA: Non-Visual Desktop Access For answers to frequently asked questions about this list visit: http://www.jaws-users.com/help/
