Most people would say that you have to be egotistical for wanting your computer to speak in your own voice, but let's think about this for a moment. Imagine losing the power of speech. It can happen and it has for a few people, some of them famous. Let's think of Stephen Hawking for example. For years, he has been using technology to get himself heard.
The main voice powering his communications system, that is now attributed to Hawking is quite familiar to most blind computer users. Known as the DecTalk speech, it was one of the most popular hardware speech synthesizers in the 1980s and 90s. To find out how Stephen Hawking communicates with the world visit this blog entry at:
Back to the purpose of my blog entry; making your computer talk in your own voice. Yes, this is indeed possible. It actually has been for a couple of years now through a system called ModelTalker. Here's a quote from the web page:
The ModelTalker System is a revolutionary speech synthesis software package designed to benefit people who are losing or who have already lost their ability to speak. It allows people with ALS or other conditions to use a synthetic version of their own voice for communication, or to choose a voice best suited to represent them.
What. You need convincing? Why not listen to my virtual voice reading "Green Eggs And Ham".
Yeah, it weirds me out too, but it shows that it works and this can benefit many people who are in danger of losing their voice. Now here's the cool thing. The final generated voice is SAPI-compliant. Meaning that you can access the voice via your favourite screen reader or speech-enabled program.
So, why would I go through the process of creating my own computerized voice? Because I can, that's why. And, also because I can help the project iron out the kinks and bugs within their system. I've also admitted at being a nerd in previous Blog entries and this is indeed a nerdy Endeavour.
What you need for this project:
First off, you need a good chunk of time. The process can be quite time consuming. The system needs to have examples of your speech sounds and will get you to read all sorts of things. Last time I went through the process, it took me a good 8 hours of speaking nonsense to my PC. The nice thing is you can stop the process at any time and start it where you left off. As for equipment, you'll need a headset and a microphone; preferably built-in to the headset so you don't need to fiddle with too many things. You'll also need a quiet space with preferably no unwanted effects such as echo. In other words, don't record this in your washroom.
Finally, you'll need the software. You need to register before you get it though. This is in order to track the use of the system and also to be able to retrieve your voice files when the process is complete. The register page is here:
Once registered, you'll need to download the Combined MTVR and ModelTalker TTS package [46 MB] and the Full Female Voice (ModelTalker Kate) [171 MB]. Install both products. I would suggest restarting your system before you try running any of the products. I had to do that to get mine to cooperate.
Now for the fun stuff:
Here are a few pointers for setting up your voice project. Two new icons have been added to your desktop. They are ModelTalker and MTVR. The ModelTalker program will let you test the voice. Run this program and type whatever you like. Press ALT+S to have the text read by Kate, the default voice. You can change her personality by pressing ALT+G for Gloom, ALT+A for anger, ALT+C for cheer and ALT+N for contradiction. For those with more curiosity, you can play with a multitude of settings under the options menu and TTS configuration. Once you've created and installed your own voice, you'll be able to control your own emotions. How cool is that?
The MTVR program is where all the action happens. MTVR stands for ModelTalker Voice Recorder. There are a few steps which you'll need to do on first run. There will be an informational dialog box which you can choose to hide or you can just close it by pressing TAB and then ENTER on the close button.
Before we can do anything, we need to create a new inventory. This is where all of your voice snippets will be saved. Press CONTROL+N to do this.
The system will ask you for your Username and Password. Type in your username, TAB to the password field, type it in and press ENTER.
A new dialog box will appear, but may not be recognized by your screen reader. The system is asking you for an inventory name. Type in something that is easy for you to remember and press ENTER.
Yet another dialog box will show up, but this one will actually read to you with your screen reader. Press ENTER on the OK button.
We're finally at the recording stage. Press the UP and DOWN ARROWS to investigate the audio prompts. Once you've found the first one, press the SPACEBAR to record the audio prompt in your own voice. A calibration dialog will show up in order to make sure that your microphone is set up correctly. TAB to the help explanation window to find out what you need to do. This is a readable window by screen readers. Tab a few times to the Start button and press SPACEBAR.
to start the process. Follow the instructions until you get back to the main screen. From then on, you'll be able to press SPACEBAR to start and stop recording. You'll need to do the calibration process every time you get back into the program and start recording again.
To hear your recording, Press CONTROL+R. To hear the audio prompt again, press CONTROL+P. Once you've successfully recorded a phrase, the system will automatically move to the next audio prompt.
To monitor your progress, check the status bar found at the bottom of the screen.
You can stop the process at any time, come back to it and the system will remember where you left off.
The first set of prompts is composed of 10 audio prompts. Once completed, the system will send it to the ModelTalker server for analysis. You will be notified within a couple of days when you can continue and what you may need to change in order to improve the sound quality.
Well, this gives you enough information for you to start the creation of your own virtual voice. Next I'll show you how to convert the rest of your body into binary code so that you can infiltrate the internet and take over the world. Meanwhile, however, I have listed some shortcuts which will ease the use of MTVR. See Below:
First audio line: CONTROL+F
Next Audio Line: DOWN ARROW
Previous Audio Line: UP ARROW
Last Audio Line: CONTROL+L
Play Prompt: CONTROL+P
Play Your Recording: CONTROL+R
New Inventory: CONTROL+N
Open Inventory: CONTROL+O
Close Inventory: CONTROL+W