You need to treat this in two stages.
The ring detect is easy, there are many schematics on the web but basically you want to use the AC ringing voltage to light the LED in an opto-coupler. The output of the opto-coupler then has to be read by a microprocessor so it can count the five rings to see if it's time to read the tones.
The tones are read by the same device as used in normal DTMF detector circuits. I normally use MV8870 devices but there are many alternatives. These ICs have a pin that tells you a tone is present and four pins that tell you what the tone number is. The microprocessor can use these to decode and verify the number being entered.
If you want to make a voice announcement to instruct the user to enter their number, you can either do it with a tape recording or far better, use a voice record/playback IC.
Brian.