Are there easy-to-use free or cheap speech synthesis libraries for embedded PIC and / or ARM systems where code size is more important than speech quality? Currently, it seems that the 1 mega package is considered "compact", but many microcontrollers are smaller than this. Back in 1980, Apple hired a Macintalk manufacturing contractor who offered reasonable, high-quality speech in the 26K package, which worked at 7.16 MHz 68000, and a program called SAM could deliver a speech that was not quite so good, but still working. , with a 16K package that worked at a frequency of 1 MHz 6502. SpeakJet runs a speech synthesis algorithm on some types of PIC.
I probably would not really need a speech production, but I would like to be able to speak messages formed from a series of predefined words. Obviously, it would be possible to simply record all the messages, but with vocabulary, for example. 100 words, I would think that storing code at 16 KB plus maybe 1 thousand phonetic lines would be more compact than storing sound at 100 words.
Alternatively, if I wanted to keep audio to 100 words, what would be the best way to generate a set of words that would naturally flow together? In old-style speech synthesizers, any word can be pronounced in three ways: a neutral inflection, a falling inflection (as if a period had followed), or an increasing inflection (followed by a question mark). Neutral kink words can be spliced together in any order and sound good. The text-wave tool that I found seems to add finer inflection details that sound “off” if the words are cropped and reinstalled. Are there any tools designed to create waves that can be combined and well fused? If I use such a tool, which audio format is best suited for storing waves in order to provide efficient decoding on a small microcontroller?
source share