Contentrue builds on third-party software, models and services. Those we are required to credit are listed here with their licences.
Turkish voices
The built-in Turkish narration voices run on our own servers. All of them are synthetic: no real person's voice is imitated. Videos that use generated narration are labelled as AI-generated.
- EMA Lightning, the speech model behind the "Ema" voice. Apache License 2.0. github.com/canberk7/ema-lightning
- Antalia-2 Mini, the speech model behind the "Antalia" voice. Apache License 2.0. pypi.org/project/antalia-mini
- Ahenk Cep by Özgür Güler, the model behind the "Ahenk Cep" voice (a designed female voice). Model weights are licensed under Creative Commons Attribution 4.0 (CC BY 4.0); the inference code is Apache License 2.0. No changes were made to the weights. Source: Ahenk (ozguradmin/ahenk-turkish-tts), weights at huggingface.co/ozgurguler/ahenk-cep-tr.
- trnorm by Özgür Güler, Turkish text normalisation that turns numbers, dates and abbreviations into spoken form. Apache License 2.0. github.com/ozguradmin/ahenk-turkish-tts