I've done this using MSs speech API in C#. When I get a minute I'll post some source code.

This works. The audio is very soft so it needs to be boosted, but maybe one of you guys can figure that out and repost a better example.

private void Say(string text)
{
try
{
_rovioAudioSocket = new Socket(AddressFamily.InterNetwork, SocketType.Stream, ProtocolType.Tcp);
System.Net.IPEndPoint remoteEndPoint = new System.Net.IPEndPoint(System.Net.IPAddress.Parse(_rovioIPAddress), _rovioAudioPort);
_rovioAudioSocket.Connect(remoteEndPoint);
textBox1.AppendText(DateTime.Now.ToString() + " : Audio socket connected\r\n");
// Prepare the audio port for sending audio from the PC back to the Rovio
string temp = "POST /GetAudio.cgi HTTP/1.1\r\n" + "User-Agent: AudioAgent\r\n" + "Host: " + _rovioIPAddress + "\r\n" + "Content-Length: 2147483647\r\n" + "Cache-Control: no-cache\r\n" + "\r\n";
szSend = StringToByteArray(temp);
//REVIEW: skipping Credentials
if (_rovioAudioSocket.Send(szSend) == -1)
{
textBox1.AppendText(DateTime.Now.ToString() + " : Error sending initial packet\r\n");
}
Stream stream = new MemoryStream();
byte[] format = new byte[0];
speaker.SetOutputToAudioStream(stream, new SpeechAudioFormatInfo(8000, AudioBitsPerSample.Sixteen, AudioChannel.Mono));
speaker.Volume = 100;
speaker.Rate = -8;
speaker.Speak(text);
stream.Seek(0, SeekOrigin.Begin);
byte[] output = new byte[stream.Length];
stream.Read(output, 0, (int)stream.Length);
_rovioAudioSocket.Send(output);
_rovioAudioSocket.Close();
textBox1.AppendText(DateTime.Now.ToString() + " : Audio socket closed\r\n");
}
catch (Exception ex)
{
textBox1.AppendText(DateTime.Now + " :: " + ex.Message.ToString() + "\r\n");
}
}

hi
can you tell me please what is audio port ? or what is it's value ?

I would greatly appreciate if someone gave an example of using GetAudio.cgi in python.