Tags: tts

Tacotron2 and Cherokee TTS

2020-11-05 Michael Conrad  3 minute read

I've abandoned trying to use the espeak-ng speech synthesizer. After spending many weeks working it, it still will not correctly apply stress rules (adding stress to unstressed words) and has other issues I've run into that are very problematic, such as where "d", "g", "n" sound so similar that it is easy to mishear these consonants.

Instead I've switched to using a Tacotron 2 TTS system published by Tomáš Nekvinda for speaking Cherokee. While more work definitely needs done, I think this is a good start, and generally sounds much better than espeak-ng. Samples follow.

espeak-ng and CED phonetics

2020-06-29 Michael Conrad  14 minute read

(Audio updated 2020-06-29)

I've been working on adding Cherokee support to the espeak-ng speech synthesizer. While more work definitely needs done, I think this is a good start.

I have chosen to use the Cherokee-English Dictionary phonetics, as this is the only written form that indicates both tone and cadence that is in common use. Hopefully the results will be good enough at a future date to be able to add pronunciations for all the C.E.D. and Raven Rock entries at the Cherokee Dictionary Project

I have posted here the espeak-ng output for 100 of the conjugated challenges from the Bound Pronouns app for review by those interested.

As a student of the Cherokee language, and not having any first language or second language speakers near by presents a challenge when working on learning the basic pitch tones of the language and the cadence of words.

Fortunately the Cherokee-English Dictionary has fully annotated pronunciation entries available.

The following examples demonstrate why creating these audio files is so important: