Adding Speech to your .NET application

It is not easy to find good and free Text to Speech libraries for your .NET application. Even the solution in this MSDN blog, is of low-quality.

By far, the best TTS solution I have come across is AT&T Text-To-Speech Project. This is a project of AT&T Research. It is actually web-based and it is free. One should however not abuse the service and make a reasonable amount of calls to it.

The source code for consuming the service is presented below. It makes a simple HTTP POST request with the text and type of voice to use as input and returns the URL of the WAV file as a string.

private string speak(string theText, string theVoice)
{
try
{

string uriString = "http://192.20.225.55/tts/cgi-bin/nph-talk";

/* Create a new WebClient instance.*/
WebClient myWebClient = new WebClient();

/* Create a new NameValueCollection instance to hold some custom parameters to be posted to the URL.*/
System.Collections.Specialized.NameValueCollection myNameValueCollection = new System.Collections.Specialized.NameValueCollection();

string voice=theVoice;
string txt = theText;
string downloadButton = "DOWNLOAD";

/* Add necessary parameter/value pairs to the name/value container.*/

myNameValueCollection.Add("voice", voice);
myNameValueCollection.Add("txt", txt);
myNameValueCollection.Add("downloadButton", downloadButton);

/* Upload the NameValueCollection.*/
byte[] responseArray = myWebClient.UploadValues(uriString, "POST", myNameValueCollection);
string response = Encoding.ASCII.GetString(responseArray);

/*This code retrieves hyperlinks from a page*/
PODN.Net.Search.PageParser pp = new PODN.Net.Search.PageParser();
PODN.Net.Search.PageLink[] links = pp.RetrieveHyperlinks(response, uriString);

string fileToPlay=null;
if(links!=null && links.Length>0)
fileToPlay = "http://192.20.225.55" + links[0].URL;

return fileToPlay;

}

catch (Exception e) { /* ERROR */}
}

The PODN library is self-made and you might well replace it with your own. Effectively is used here to retrieve the hyperlinks from a web page.

Later you can add a Windows Media Player control in your application and have it play the audio file. Your computer has now voice! As an example app you could download news and let the program read them back to you. You can find more details and the example application ('Tell me the news') in my 'Adding Speech to .NET applications' blog post here.

The AT&T tts project can be found here.

A funny detail, taken from the project website:

"...This movie, starring Sean Penn and Michelle Pfeiffer ("My Name is Sam"), used our "Crystal" as the voice of Pfeiffer's voice-activated cell phone. Please note that we take no responsiblity for the infuriatingly inaccurate speech recognition in the movie "

The System.Speech.Recognition is nice in the WPF but the System.Speech.Synthesis I whink you will agree, is of inferior quality to the voices in AT&T research tts project.

One interesting idea, though,  that is described in the blog link you provided is to read text from e-books that can be found in the Gutenberg Project. Nice!

WPF supports text-to-speech in System.Speech.Recognition. It sounds though like you are already unimpressed with Windows' TTS capabilities.

EDIT:  Oops, I meant System.Speech.Synthesis.  Thanks for the catch.  Also, the speech recognition and synthesis is a .NET 3.0 feature not necessarily tied to WPF.

Hi, I checked your site and I downloaded your text-to-speech example application here. Can you give the source code for your PODN library that you use? I am afraid it has some bugs when reading the news (it says only 3 instead of 5 news)
thanx

Hi Dan,

I am afraid I haven't and I am sorry that I didn't mention that the speech solution in the post can also be run on older versions of the .NET framework. All you need is a decent internet connection. No libs and no special software while the result is really of high quality.

Unfortunately, I have not yet migrated to Vista or to new .NET Frameworks or even to Microsoft VS 2008 that you are referring. Some unfortunate incidents like this Windows Vista Speech product presentation, made me postpone the transition and wait for things to stabilize!

Just kidding! I am sure it is great..

Spam!

what's wrong with using microsoft's text to speech API?  It's free, works with .net 2.0, and for a bit of money you can create your on synth voice?  I've been using it for a while to verbalize our errors (since the color changing status bar with text in it was tasking some users for comprehension)..

Hi modi123,

I didn't say that Microsoft's tts is a bad solution. I have used it myself for personal projects too.  However there is a notable quality difference between the two solutions. At least this is my opinion.