How I use Dragon Voice to write lyrics
Hello, my name is Mathieu Lalonde, from Mathieu Media Production. I use Dragon Voice software to write. I am writing this text to give you some tips to help you improve your use of speech recognition software for people with disabilities.
About me
Let’s first say that I suffer from cerebral palsy. Cerebral palsy can affect several parts of the body. In my case, my legs, fine motor skills, and my arms are affected.
Before the introduction of speech recognition
Before using speech recognition software, I typed my texts using only two fingers. So, writing a paragraph of 12 to 15 lines took me at least three to six hours. It was very long and sometimes discouraging. It took me two to three days to write a letter. I wondered if speech recognition software could help me write faster.
Why use speech recognition software
- It’s easier for me, because of my fine motor skills, to type my texts on the computer than to write them by hand.
- To communicate more quickly with Mathieu Média Production customers
- To gain speed in writing my texts
- To enable me to have easy-to-reuse business documents
In 2015, I discovered a software that partially meets my needs for writing my texts. Thanks to funds from the Phénix as part of the Young Entrepreneurs with Disabilities program, I was able to get my hands on the Dragon Voice voice recognition software.
My beginnings with Dragon Voice software
I thought the voice recognition software would have a hard time understanding what I was saying, but I tried it anyway.
I installed the speech recognition software from the CD and followed the steps on the screen. Then I created a profile so that the software would adapt to the tone of my voice and the way I speak. Then I read the training texts so that it would adapt to my voice. I started reading the training texts and I found it very difficult, but I told myself not to get discouraged. I told myself that the texts that the speech recognition software asked me to read were texts predetermined by software to better evaluate my voice.
My first use of the speech recognition software was a great success. It took me 48 minutes to write a three and a half page email. Normally it would take me three to four days to write the same email. At first, the software recognized my voice well, but I noticed a decrease in efficiency when I used it within word processing or when filling out forms.
Decreased voice recognition and computer problems
The speech recognition software was opening unwanted software on my computer and this was causing my computer to freeze.
I wondered why the speech recognition software wasn’t working well for me. I called Phoenix to discuss my difficulties writing with the speech recognition software and from there, we started doing some research to understand why the software was having difficulty recognizing my voice. Here are the questions we asked ourselves.
Question 1: Was my audio equipment adjusted optimally for the speech recognition software?
I thought my audio signal was of poor quality and I thought it might help if I used better quality audio equipment. I checked the audio first because I am an entrepreneur in the audiovisual field. I started using the professional microphone and a sound console to improve the quality of the audio that the speech recognition software receives.
I experimented with various frequencies, starting with very low ones, but it didn’t work well. The best position seemed to be the middle one but the software still had difficulty writing what I dictated.
Question 2: Could my microphone volume be causing the problem?
I thought about the problem and decided to try with my professional microphone and a sound console to improve the audio quality. I started by adjusting the sound of my computer (Windows XP operating system) so that I could hear my sound console which is connected to the line input of my computer’s sound card. Then, I turned the volume of the line input to maximum so that I could hear my sound console. I did a first test to see if my audio was loud enough for the speech recognition software to transcribe my words better but the sound was too loud. So I started to lower the sound on my console but it was too low for the speech recognition software. Then, I increased the volume of my console little by little until I found the ideal volume for my professional microphone. And I noticed that the software recognized my voice better but still made typing errors in the word processing software.
Then I asked myself the following question.
Question 3: Could my pronunciation contribute to the problem?
This would be possible since, due to my cerebral palsy, there are certain words that I have difficulty articulating properly. To verify my hypothesis, the Phoenix put me in contact with the Moi, j’apprends center.
We organized a first preparatory meeting. The trainers wanted to see how the voice recognition software worked and assess how the Moi, j’apprends center could help me.
This first meeting took place via Skype, in a conference with the Moi, j’apprends center, the Phénix counselor and myself. We talked about my physical difficulties and my needs. Together we determined that it was important to check these three points:
- Do I pronounce my words clearly?
- Am I speaking too fast?
- Does the software recognize the French-Canadian accent?
Practice diction with speech recognition
I started diction lessons with the collaboration of Moi, j’apprends so that they can determine if my language is difficult to understand because of my cerebral palsy.
First, we assessed the quality of my spoken French. During our conversation practice sessions, we found that another person has no difficulty understanding what I say.
Next, we researched the speech recognition software but couldn’t find much information. We created a new user profile by following the instructions exactly at each step. After creating the new profile, I started reading texts that weren’t in the texts provided by the speech recognition software. The Moi j’apprends trainer sent me texts that she had selected for me to practice on to see how my voice would be recognized with a new profile. She also asked me to record myself and also to read a single text twice a day and record them both times so she could assess if there was a difference in the tone or rhythm of my voice between the morning and the afternoon.
For a few days, I read a single text in the morning and then in the afternoon I sent an email to the trainer, with a video link and also a PDF. By watching my videos, she noticed that my words were understood very well. On the other hand, there seemed to be a difficulty in recognizing small words like le, la, les, des, dans, cet, cette, comment, quand.
In the third meeting, we practiced these words. Also, I tried to pronounce my words “like a French person from France” but it didn’t help. So we continued to practice together to find out how I should adjust so that the speech recognition software could hear my voice better to write better.
In the fourth meeting we explored the speed at which I read to see if I should speak very quickly, very slowly or like a robot.
We started with a reading test with a normal flow to have a comparison with the other exercises that I would do. We noted several errors in the speech recognition.
Then I read the same text again but faster, almost as if I were at an auction! There were still mistakes but not in the same words. Finally, I read the same text again but this time at a slower reading speed. After reading this text, we noticed that the speech recognition software was more effective if I spoke slowly.
That’s when I wondered if it would be even better if I spoke a little like a robot, that is, pausing between each word and clearly articulating each word. I took the same text again and we noticed an almost error-free recognition of my text.
We then said to ourselves that now we had found the speed at which the voice recognition software best recognized my voice. And that this speed also allowed the computer to follow me better.
Other questions
Why did word processors freeze or nearly freeze my computer while I was using speech recognition software?
When I was writing text in a word processing software, sometimes the speech recognition software might freeze my computer. So I decided to try different word processing software.
Open Office
I tried using the Open Office word processing software with the speech recognition software. It worked very well sometimes but also very poorly sometimes. Open Office would start two or three other programs at the same time, which could cause my computer to freeze.
I tried to use Open Office to answer a questionnaire but it didn’t work well: 150 programs started at the same time without me asking. It almost froze my computer and it took at least 10 to 15 minutes before I could start writing my answers again. I deduced that the Open Office word processing software has problems integrating with the speech recognition software.
Free Office
Then I tried Libre Office and did the same two tests again. I had a little less difficulty but this software also started some programs without me asking for it via voice recognition.
WordPad
Then I tried my luck with Microsoft Wordpad. I was surprised: the software recognized my voice better when I wrote my texts with Wordpad. So, now I use Microsoft WordPad software with the speech recognition software to write my texts.
How do I answer questions written in a word processing program?
I always try to answer directly in the questionnaire first, but if I have difficulty, I write my answers in Microsoft WordPad software and copy them into the word processing software where the questionnaire is located.
Why is speech recognition software more effective in the morning than in the afternoon?
I noticed a difference between the texts I wrote in the morning and the ones I wrote in the afternoon. I did some research on the speech recognition software’s support site. I calculated how much time I had to write my texts before the speech recognition software started making multiple errors. I realized that I only had 30 to 40 minutes to write my text before giving the computer a 20-minute break so it could clear its RAM so I could continue writing my text.
If the problems continue after the 20 minute break, I restart the computer.
Conclusion of my research
As a conclusion to my research, I realized that the Dragon Voice speech recognition software that I use is a very effective software for writing and very efficient if you use Microsoft WordPad and then I copy my writing from the Microsoft software and paste it into the word processing software of my choice.
My little tips for using
voice recognition software like Dragon Voice.
- Take your time to create your profile in your speech recognition software.
- Take the time to read the documentation provided by the
speech recognition software. - Also read the information you find on websites or discussion forums about speech recognition software.
- Take the time to dictate your text carefully.
- Close all unnecessary applications while you type your text to get the best possible performance from the speech recognition software.
- Take breaks as needed to allow the computer to clear its RAM.
- Adjust the speed of your speech if necessary.
- Pronounce the words clearly
- Adjust your microphone closer or further from your mouth as needed
- Use a headset microphone if you have difficulty holding a microphone.
- Speak louder if necessary.
- Feel free to proofread your text or ask someone else to read it.
- Use a spell checker to correct mistakes.
Discoveries with the new headset
At the time I wrote this text, I was using a microphone that I held in my hand and the speech recognition software had a lot of difficulty hearing me. But now that I use a headset microphone to dictate my texts, it’s much better.
Here are my results with the new headset
Using the headset microphone, I noticed that Dragon Voice understands me better and now I can speak faster, the software is able to follow me 90% of the time and there are almost no mistakes.
As a final conclusion to this work.
By creating this book I wanted to help people with disabilities so that they can use voice recognition software like Dragon Voice to write more quickly every day.
I also wanted to show that you should not get discouraged when you start using speech recognition software. Yes, I know from experience that it is a little discouraging but with determination you can succeed in using speech recognition software as if it were your pencil in your hand. I encourage you to try it and if you need help, do not hesitate to contact the organizations in your area: they are there to help you.
Together beyond disability!
Supporting note
For the writing of this document I used the following material:
- The voice recognition software, Dragon Voice.
- A computer with the following specifications: Windows XP operating system (OS) with a 2.4 megahertz dual-core processor with 4 GB of RAM, a 1TB hard drive
- Microsoft WordPad
If you would like more information about the speech recognition software used in this text or to give it a try, please visit their website at https://www.nuance.com/for-individuals/index.htm
Please note:
- The Internet links provided in this document maychange at any time without notice.
- I tested on Windows XP. Results may vary from system to system.
- You assume the risk: I assume no responsibility for the information provided in this information sheet.
Thanks
I would like to thank the Phoenix and the Moi, j’apprends Adult Training Center for helping me do my research to write this document. If you would like to know more about the Phoenix and Moi, j’apprends, I invite you to visit their respective sites.
The Phoenix: https://www.lephenix.on.ca/