• Tip Jar

Getting More Voices For Orca

Festival Voices

In Ubuntu Orca uses Espeak by default. This is my prefered speech synthesizer, but it is not the only choice. Other Linux distributions use the Festival speech engine instead of Espeak. It is possible on Ubuntu to use festival if you want. First, you need to install Festival and some voices to use with it. Once that is done, you can use the Festival voices with Orca by opening the Orca preferences, Orca+space bar, press right arrow to get to the speech tab, and tab over to the speech synthesizer combobox. use up and down arrows to move through the different synthesizers, Festival Gnome speech driver should be one of the options. To get Festival and the 16K versions of the Festvox Diphone voices, type:
sudo apt-get install festival festvox-don festvox-kallpc16k festvox-kdlpc16k festvox-rablpc16k
If you would rather have the 8K versions of the above voices for some reason, replace 16k with 8k when installing them. These default voices aren’t the best festival has to offer. There are other voices that sound better. There is an excellent guide for installing the Mbrola, cmu, and Nitech voices on the Ubuntu Forums. If you are using the 64 bit version of Ubuntu, you will need to first install the ia32-libs package before installing Mbrola with the .deb package. In terminal type:
sudo apt-get install ia32-libs
Then, in the instructions, watch for the line that starts with sudo dpkg -i. After the -i, type –force-architecture and then complete the command. So, the whole command should look like this:
sudo dpkg -i –force-architecture mbrola3.0.1h_i386.deb
If you are not using the 64 bit version of Ubuntu, you do not have to worry about this step, just follow the instructions as written.
I have created an application that will install the voices you select if you would rather not have to type a lot of things in the terminal. You can use it to install Mbrola and Nitech voices. It is located at:
http://www.stormdragon.tk/voices-script/
If you are using 64 bit Ubuntu, you will still need to install the ia32-libs package before running the script.

Purchasing Voices

The IBM Viavoice is available from Voxin for a small fee, about $5.75 USD. They They are the Eloquence voices. They depend on an old library that hasn’t been updated in years, so it is possible that they may quit working on future releases of your distribution. The purchase proceedure is quite simple, only one language comes with your purchase, so if you want multiple languages, you will have to purchase multiple packages. These voices are not sold for profit. The newest version now supports 64 bit systems.
Cepstral offers voices for purchase for both 32 and 64 bit systems. To install them, simply unzip the package you have downloaded, open terminal and switch to the directory you have just unzipped, and type:
sudo ./install.sh
Here’s a good place to point out the autocompletion feature in terminal. If you type part of the directory you want and press the tab key the rest will be filled in for you. It works with commands too, so give it a try, I am sure you will like it. To get them working with Orca, you will need to install gnome-speech-swift. To do this, in terminal, type:
sudo apt-get install gnome-speech-swift
If you run in to problems with it not showing up in Orca’s speech synthesizer list, there is more information available on the Orca wiki.
It’s always a good idea to test out a new speech engine before switching to it with Orca. There’s nothing worse than switching to a new synth just to find out there is something wrong with it and not have any speech to correct the problem. So, to test Cepstral, you can use the swift command. If you have installed it in the default location, the command you would use to test it is:
swift -n VoiceName “hello”
Replace VoiceName with the name of the voice you would like to test.

Notes

This is a work in progress. I will add more voice info as I find out about them. There are some things not covered here like Dectalk, Freetts, etc. When I have time to test/get the information I will post instructions for them as well. The commands in this post were done using Ubuntu 8.10 Intrepid. This should work on most Debian distributions. Commands for other distributions may be different.

Bookmark the permalink.

2 Comments

  1. Yes I need help I used your guide to install the voices but i do not see them in orca prefrences can you help me use them for my disabled friends at school

  2. The way voices are added to Orca is different now. Everything is handled through speech-dispatcher. After you have added the voice check for both ~/.speech-dispatcher/conf/speechd.conf and /etc/speech-dispatcher/speechd.conf and make sure the line in voice modules containing your synth’s info isn’t commented. So, for example, if you were using Cepstral, you would make sure the line that handles swift-generic did not have a # at the beginning.

Comments are closed

  • Tip Jar