There are many different solutions available nowadays for storing speech, recording it, and then playing it back. For embedded applications, it is in most cases of advantage to opt for a single chip solution. This in turn, in most instances, is the ideal possibility for integrating spoken texts, or speech output in accordance with specific processes, into one application.
There are also major differences in the configuration or format. The speech output for a greetings card is extremely cheap, and really needs no quality at all. But if a speech output is being integrated into a security system, the requirements are entirely different, and the voice must be very clearly understood.
Codico has a relatively wide range available for many application situations of this type.