NWU-MuST/ttslab2_afr_lwazi2_build
Domain:
natural language processingRecord type:
software
Creator:
NWU
Host:
TTSLab Lwazi2 Afrikaans build scripts
TTSLab Lwazi2 Afrikaans build scripts
=====================================
This repository contains pre-setup data and scripts for building an Afrikaans TTS voice using [TTSLab][1] and associated [build tools][2]. This repository should be built against TTSLab `commit 5cb979c...`. The recommended way to do this is described in `./HOWTO.md` and involves cloning this repository and running the accompanying [Docker][3] script (`./Dockerfile`).
The copyright and licence information for scripts in this repository can be found in `./COPYRIGHT` and `./LICENCE-APACHE`/`./LICENCE-MIT`. This repository also contains software and data developed by third parties used under licence in the `./src` and `./recs` directories. Copyright and licence information for third-party components is contained in each individual sub-directory or source file. Summaries of these components including attribution information are briefly presented below.
Since the speech corpus bundled here is specifically released under a licence that prohibits commercial use, this repository can be used as-is for research purposes or to demonstrate how to use the TTSLab toolkit to build voices (__in academic work please cite 1__).
## Speech corpus and language data
#### Lwazi II Afrikaans TTS Corpus
The Lwazi 2 Afrikaans text-to-speech corpus was originally developed by the CSIR in South Africa and can be used and distributed under the terms of a [Creative Commons Licence][4] which prohibits commercial use. If this corpus is used for research purposes, please refer to the CSIR technical report (__cite 1__).
For more information and links to the original corpus distribution refer to `./recs/README.md`.
#### Afrikaans Pronunciation Dictionary
The pronunciation dictionary used here was enhanced for use in TTS during the NTTS project funded by the Department of Arts and Culture (DAC) of the Government of South Africa and is [available on Github][5] under a Creative Commons licence (__cite 2 or 3__). The ori …