Is there a text-to-speech software that accepts IPA-based phonetic transcription? [closed]

Question

I would like to translate phonetic text (IPA) to synthesized speech. No TTS software that I have found can do this and I can't find any other software either. Any tip? Is this even doable in theory with decent quality or isn't the phonetic text (IPA in this case) good enough for this?

Loquendo's TTS has support for reading phonetic text but the result is very bad. I'm going to contact their support to get some help.

Pate

This question is closed now, so I hope it will eventually be migrated to softwarerecs.stackexchange.com. – Walloon

Answer 1

http://ssb22.user.srcf.net/gradint/lexconvert.html might be what you need; it apparently converts between various text-to-speech software encodings, including SAMPA and IPA-Unicode; with this eSpeak can be made to pronounce the encoding.

Good luck!

Syncarpous

Wow! This works great by copying-pasting IPA from Wikipedia articles and hearing them on my Mac (using the --trymac option). This is really really cool -- considering that eSpeak has been "converted/recompiled" to JavaScript using emscripten (see here) this means a bookmarklet/chrome plugin would be more than doable. Very inspiring! – Hasbeen

Answer 2

A phonetic representation will not suffice for a good speech synthesis. The first question would be is your phonetic representation correct? There are many unsolvable or at least very expensive and complex problems when trying to create phonetic text out of normal text. To keep it simple: You would need to know the origin of each word in order to be sure how to pronounce them. Even if your phonetic representation is perfect you probably lack any information about prosody, which is also a very important part of good speech synthesis.

Segalman