Shorthand Dictation
Nonfic.DictProg History
Hide minor edits - Show changes to markup
| 140 | didn't use the program |
| 140 | without delay codes |
wpm delay 40 700 50 450 60 300 70 200 80 125 90 60 100 10, 7, 5, 1 140 didn't use the program
| wpm | delay |
| 40 | 700 |
| 50 | 450 |
| 60 | 300 |
| 70 | 200 |
| 80 | 125 |
| 90 | 60 |
| 100 | 10, 7, 5, 1 |
| 140 | didn't use the program |
I use Ceptsral's SwiftTalker.
I use Cepstral's SwiftTalker.
Face it, there is no perfect way to judge dictation speed. There are always reasons for doing it another way. So let's keep it simple.
For the most careful tests, I used paragraph 7-54 from Gregg Shorthand Manual Simplified, Second Edition. It claims to have 52 words. Okay, not a very large sample.
I used the Lawrence voice at 170 wpm.
I ran it through the program to put in breaks, then exported the audio file. I then removed the "register me" messages (using Audacity), and here are the results:
It's impossible to say which delay value will give what speed. There are too many variables.
I chose a voice, and set my goals in "delay" rather than wpm. I only convert back to wpm when talking to other humans, or when I want to increase my speed by a given amount.
This is the chart I use for converting:
As you can see, the stated speed for this voice is high, although other tests with other voices say it's right on.
For speed-building, I chose a single voice, and set goals based on delay rather than wpm, then converted when talking to others.
The chart was made as follows:
Paragraph 7-54 from Gregg Shorthand Manual Simplified, Second Edition. It claims to have 52 words. Yes, not a very large sample.
Lawrence voice at 170 wpm.
I ran it through the program to put in breaks, then exported the audio file. I then removed the "register me" messages (using Audacity), and did the math.
You'll notice that, although I told it 170wpm, it gave me 140. Other voices gave other results.
Musings
Schools count 1.4 syllables as one word, because a passage with many long words will take longer to write.
However, that doesn't give a good indication of your speed in the field.
If you're familiar with the field, you will have short forms for the longest words, so a 5-syllable word would be easier than 5 single-syllable words.
Swem himself said that cases with lots of numbers were difficult to record. You can't phrase them.
The championship speeds for literary material are much slower than for Q&A, which are slower than court proceedings. Same writers, same days. Proof that the material makes a huge difference.
Natural human speech phrases things differently than machine speech. Trust me on this one.
So, while it's useful for comparing your progress, an accurate wpm isn't really necessary this program.
If you're preparing for an exam, it's best to spend the money on their practice material for your final preparation.
Any speed over 100wpm can be done through Cepstral rather than this program.
Any speed over 100wpm can be done through Cepstral rather than this program, which saves a step.
The program replaces spaces with the "break" command. Not sure he included other whitespace or what happens if there is more than one space in a row. (He also did dishes that night, so I didn't get too picky.)
The program replaces spaces with the "break" command. Not sure what he did with other whitespace or what happens if there is more than one space in a row. (He also did dishes that night, so I didn't get too picky.)
It adds more time after periods and other punctuation. When typing, leave out the period after "Mr". (That follows modern British punctuation.)
Unless otherwise stated, I've used the following rules for calculating speeds:
"Words" is actual words, regardless of syllables or potential for brief forms and phrasing.
"Time" is the full time, starting from the first word of the dictation, ending with the beginning of the next passage.
If using a free voice, count the time spent on advertisements. (It's time to catch up.) Do not count the words in the advertisement unless you intend to write them. Do not count the time on the advertisement before the passage starts. (You aren't writing anything yet.) Yes, I consider it unfair to reduce all 10 words of the long ad to 1 brief form.
For the most careful tests, I used paragraph 7-54 from Gregg Shorthand Manual Simplified, Second Edition. It claims to have 52 words. Okay, not a very large sample.
I used the Lawrence voice at 170 wpm.
I ran it through the program to put in breaks, then exported the audio file. I then removed the "register me" messages (using Audacity), and here are the results:
wpm delay 40 700 50 450 60 300 70 200 80 125 90 60 100 10, 7, 5, 1 140 didn't use the program
As you can see, the stated speed for this voice is high, although other tests with other voices say it's right on.
For speed-building, I chose a single voice, and set goals based on delay rather than wpm, then converted when talking to others.
My own experiments used a piece 70 words long, 89 syllables, for an SI of 1.27. Cepstral was set at 170wpm.
My first experiments used a piece 70 words long, 89 syllables, for an SI of 1.27. Cepstral was set at 170wpm.
Unless otherwise stated:
Unless otherwise stated, I've used the following rules for calculating speeds:
If using a free voice, count the time spent on advertisements, but do not count those words (unless you actually write them). Do not count the time on the advertisement before the passage starts. (Yes, I consider it unfair to reduce all 10 words of the long ad to 1 brief form.)
If using a free voice, count the time spent on advertisements. (It's time to catch up.) Do not count the words in the advertisement unless you intend to write them. Do not count the time on the advertisement before the passage starts. (You aren't writing anything yet.) Yes, I consider it unfair to reduce all 10 words of the long ad to 1 brief form.
- rabbit300.wav
- Delay of 300ms between words.
- rabbit1000.wav
- Delay of 1000ms between words.
- rabbit300.wav
- Delay of 300ms between words. File is over 3MB.
- rabbit1000.wav
- Delay of 1000ms between words. File is just under 1MB.
- Rabbit300.wav Δ
- Delay of 300ms between words.
- Rabbit1000.wav Δ
- Delay of 1000ms between words.
- rabbit300.wav
- Delay of 300ms between words.
- rabbit1000.wav
- Delay of 1000ms between words.
Sample .WAV, Summary
Here are two sample .wav files, using Cepstral's David. They are the same passage. The slower file was created at a lower frequency/quality, hence the drastic difference in file size.
- Rabbit300.wav Δ
- Delay of 300ms between words.
- Rabbit1000.wav Δ
- Delay of 1000ms between words.
http://onebit.ca/cgi-bin/short.py
http://onebit.ca/cgi-bin/short.py
Delay wpm
10 110
50 100
120 90
200 80
300 70
440 60
| Delay | wpm |
| 10 | 110 |
| 50 | 100 |
| 120 | 90 |
| 200 | 80 |
| 300 | 70 |
| 440 | 60 |
| border |
Delay wpm 10 110 50 100 120 90 200 80 300 70
Delay wpm
10 110
50 100
120 90
200 80
300 70\\
The formula to calculate any speed using Lawrence based on these results is y = 107.9107e ^ 0.0013902x, where y = wpm and x = delay in milliseconds, so you need to be pretty hot at maths to get other speeds, I'm afraid.
The formula to calculate any speed using Lawrence based on these results is y = 107.9107e ^ 0.0013902x, where y = wpm and x = delay in milliseconds, so you need to be pretty hot at maths to get other speeds, I'm afraid.
Hi Cricket,
I love your little program and have used it with Cepstral voice 'Lawrence' (being British) purchased, so without the inserted nags. My results are a bit different from yours but maybe Lawrence speaks more slowly(!). I went for an SI of 1.30 and experimented with various delays to get a 2 minute speech at different wpm. Results were as follows:
Delay wpm 10 110 50 100 120 90 200 80 300 70 440 60 The formula to calculate any speed using Lawrence based on these results is y = 107.9107e ^ 0.0013902x, where y = wpm and x = delay in milliseconds, so you need to be pretty hot at maths to get other speeds, I'm afraid.
Good luck with it - it's a really useful tool for shorthand learners.
(:nl:)>>messagehead<<
Cricket? — 10 October 2007, 12:06
http://onebit.ca/cgi-bin/short.py
http://onebit.ca/cgi-bin/short.py
http://onebit.ca/cgi-bin/short.py
http://onebit.ca/cgi-bin/short.py
Shorthand is a hobby of mine. Unfortunately, dictating at the slow speeds I need is not a hobby of anyone I know.
Test Comment
More test results:
Again using the free David voice, not including the time for the initial "buy me" bit. Including the time for the ads after that, but not counting those words. SwiftTalker set for 170 wpm, the default. Using all the text in chapter 2 Anni, about 768 words. A word is a word; I didn't count syllables.
| border | |
| Delay | WPM |
| 1600 | 20 |
| 1250 | 33 |
| 900 | 40 |
| 650 | 50 |
| 450 | 60 |
| 300 | 70 |
''If you experiment, please add your observations in the comment field. I will merge them into the text as I have time.
If you experiment, please add your observations in the comment field. I will merge them into the text as I have time.
Phrasing
You can omit the break between two words by replacing the space with a dash. The two words will be said together. This will reduce the total dictation time.
More Details
Speeds
Speed Results, Summary
Speeds, Early Experiments
Comments
Quick Instructions
Quick Instructions
!!Quick Instructions
Quick Instructions
Quick Instructions
Quick Instructions
The free David voice at a delay of 750 gives 45-50 wpm.
The free David voice at a delay of 750 and wpm of 170 (the default) gives 45-50 wpm.
This is where it gets frustrating.
There's an ongoing debate whether you should count syllables or actual words when bragging about your dictation speed. (Standard for one system is 1.4 syllables per word, another insists on actual words.) This is in addition to the differences between familiar and unfamiliar subjects and how much repetition there is.
Anyhoo, Cepstral's own wpm seems to use 1.4 syllables/word. (Thanks sidhetaba for those experiments.)
Face it, there is no perfect way to judge dictation speed. There are always reasons for doing it another way. So let's keep it simple.
Unless otherwise stated:
"Words" is actual words, regardless of syllables or potential for brief forms and phrasing.
"Time" is the full time, starting from the first word of the dictation, ending with the beginning of the next passage.
If using a free voice, count the time spent on advertisements, but do not count those words (unless you actually write them). Do not count the time on the advertisement before the passage starts. (Yes, I consider it unfair to reduce all 10 words of the long ad to 1 brief form.)
Speed Results, Summary
Cepstral's own wpm seems to use 1.4 syllables/word. (Thanks sidhetaba for those experiments.)
The free David voice at a delay of 750 gives 45-50 wpm.
''If you experiment, please add your observations in the comment field. I will merge them into the text as I have time.
Speeds, Early Experiments
testing comments
I did the first three units of Gregg Anniversary. A delay of 750 and speed of 170 gave 42 to 47 wpm. Many short words, so low SI.
Any speed over 100wpm can be done through Cepstral rather than this program.
Feedback is much appreciated! See the link on the sidebar.
If anyone can fit a formula showing words, syllables (or SI), and delay to get a given WPM, I'd appreciate it.
Feedback is wanted. If you want to email rather than comment here, see the sidebar.
(:commentbox:)
My own experiments used a piece 70 words long, 89 syllables, for an SI of 1.27. I subtracted the time spent on Cepstral's advertisement.
My own experiments used a piece 70 words long, 89 syllables, for an SI of 1.27. Cepstral was set at 170wpm.
| delay | net time | wpm | spm |
| delay | net time | wpm | spm |
| delay | net time | wpm | spm |
| delay | net time | wpm | spm |
(:title Shorthand Dictation:)
Hubby has written an online utility which, combined with a (free) voice synthesizer, will dictate text at the speed you choose.
It does this by adding time between the words rather than slowing down the words. (Slowing down the words too much becomes incomprehensible.)
http://onebit.ca/cgi-bin/short.py
Functional rather than fancy, but reliable. (I notice he's even put in code for invalid entries, but haven't tested it.)
Quick Instructions
Enter the text on the page as the form says. Hit "Translate".
Copy the resulting text into a text-to-speech program such as Cepstral. It has to recognize SSML (Speech Synthesis Markup Language, similar to HTML and XML. More here: http://www.cepstral.com/cgi-bin/support?page=ssml )
Press "Play."
More Details
The program replaces spaces with the "break" command. Not sure he included other whitespace or what happens if there is more than one space in a row. (He also did dishes that night, so I didn't get too picky.)
I use Ceptsral's SwiftTalker. http://www.cepstral.com/downloads/
They have an online demo at http://www.cepstral.com/demos/ that sends you a wav file.
To get SwiftTalker, download and install any of their free voices; it gets installed with the voice. If using a free voice, it sticks "buy me" in every now and then.
SwiftTalker will play it immediately or save to a wav file. (File / Export) Audacity or Nero should be able to convert it to MP3 or a CD or whatever.
You may have to go into "tools / options / text handling " so it handles SSML. You can also play around with the speed in WPM. Lots of things to play with.
Speeds
This is where it gets frustrating.
There's an ongoing debate whether you should count syllables or actual words when bragging about your dictation speed. (Standard for one system is 1.4 syllables per word, another insists on actual words.) This is in addition to the differences between familiar and unfamiliar subjects and how much repetition there is.
Anyhoo, Cepstral's own wpm seems to use 1.4 syllables/word. (Thanks sidhetaba for those experiments.)
My own experiments used a piece 70 words long, 89 syllables, for an SI of 1.27. I subtracted the time spent on Cepstral's advertisement.
| delay | net time | wpm | spm |
| 1 | 24 | 175 | 223 |
| 100 | 40 | 111 | 141 |
| 150 | 43 | 100 | 127 |
| 200 | 54 | 93 | 119 |
| 250 | 49 | 88 | 111 |
| 500 | 73 | 65 | 82 |
| 700 | 87 | 53 | 68 |
| 750 | 84 | 51 | 64 |
| 1000 | 102 | 42 | 53 |
| 1250 | 119 | 36 | 45 |
| 1500 | 140 | 32 | 40 |
| 1750 | 164 | 27 | 34 |
Feedback is much appreciated! See the link on the sidebar.