On this page you can download binary and source code releases of FaceSpeaker. This is an Experimental product provided "as is" under the terms of the GNU/GPL 3.0 license. By acquiring, installing or using (part of) FaceSpeaker you accept the license terms as outlined in the GNU/GPL 3.0 license agreement. While trying FaceSpeaker cannot normally harm your computer and you may fully uninstall the product at any time, trying FaceSpeaker is only recommended for users having reasonable computer skills and experience installing software. The FaceSpeaker documentation assumes the user has reasonable computer skills.
- Modern computer featuring at least a quad core CPU and 4GB of random access memory. FaceSpeaker may run on older hardware but this has not been tested and performance will probably be poor.
- An X64 (64 bit) edition of the Microsoft Windows 7 or the Microsoft Windows 8 operating system. Support for x86 (32 bit) operating systems is not planned because those operating systems ordinarily should not be installed on modern computers meeting FaceSpeaker's CPU and memory requirements. However, it is possible to compile an X86 release if there is significant demand for it.
- A webcam or an analogue camera combined with a video capture device (not all such devices will work). Webcams integrated into a monitor or laptop will usually work. It is not currently possible to choose which webcam or capture device to use, so connect one at a time or temporarily disable the secondary camera in device manager if FaceSpeaker uses the wrong camera.
- The .net framework 4.0 must be installed and a properly configured SAPI 5 speech engine must be present. Both conditions are normally met in a Windows 7 or Windows 8 installation but double check those things are properly configured if you run into problems.
You may download the latest version of FaceSpeaker below. This is a .msi (Windows installer) file. Open it and follow the standard installation wizard to install FaceSpeaker. A FaceSpeaker shortcut will be added to the programs / apps section of your start menu. You can completely remove FaceSpeaker using the "programs and features" control panel item, note that this Will also delete the database of trained faces! The FaceSpeaker database (containing your trained faces) is located in c:\users\yourname\appdata\roaming\facespeaker\facespeaker.sdf.
Please remember to contact the author if you have any feedback or suggestions for improvement!
If you run into a problem, double check your system meets FaceSpeaker's system requirements. Try to interpret and resolve the error message (if any) you see. You may also want to try a different webcam if available. If you get stuck don't hesitate to contact the author, by doing so you help make FaceSpeaker better. In your message, provide a precise description of the problem you experience and include the text of the error message (if any). Also provide your Windows version and the type of webcam / video capture device you are using.
The source code still needs to be cleaned up and documented better. Be sure to read the below comments and don't hecitate to contact the author if you have any questions or comments.
"Readme" (really just a message sent to someone interested in the code - still need to share the code in a more decent manner): This is a visual studio 2010 C# project. I'm running visual studio ultimate edition, it would probably also work in the free express edition but I haven't tested.
This source code is newer than the binary version which can be downloaded from the website, and the database format has changed. If you had installed the old version, you should remove the FaceSpeaker database which is located in C:\users\username\AppData\Roaming\FaceSpeaker The program first attempts to find the facespeaker.sdf database file in this location, if it is not present it will use the facespeaker.sdf included with the source package (which is copied to the bin directory upon project build). So if you want to experiment and be sure you don't lose the database, you should copy the facespeaker.sdf file included with the source to C:\users\username\AppData\Roaming\FaceSpeaker.
The program is an X64 version, as explained on the website it wouldn't make much sense to run this on an X86 platform but if that's what you need you would have to replace the emguCV files and change the architecture (a bit involved - see the emguCV wiki or I could help you get it done) The face recognition code is in OcvFaceEngine.cs, the interfaces (IFaceEngine) contain most documentation. You could "plug in" another face engine by implementing the IFaceEngine interface and some related interfaces.
Note that I didn't use the emguCV webcam functionality, this was not stable enough - I used a different open source webcam library. The user interface contains some "self voicing" functionality for blind users but this is not yet implemented that well, when you press keys while the program runs expect to hear some jibberish.