French speech as part of the final set of demo videos for the YourTTS model that I've been training. The first two voices are native French speakers, and the third voice heard is a cross-language speaker (voice trained in German, speaking in French)
This model has trained for approximately 1810000 steps on a varying assortment of the VCTK dataset, LJSpeech, CSS10, and some LibriVoice books.
This is a multilingual model, with the following languages trained: English (UK), English (American), Spanish (European), Spanish (Latin American), French, German. The following accents were trained in as separate Languages and can be applied to other voices: Irish accented English (en-ir), Indian accented English (en-in)
I'll post the model for download in the next day or so. Just need to test a couple things. All of the voices aren't perfect, but it could be a good base for continued training, and some of them do sound pretty good. Worth trying out the cross language ability; have a speaker talk in a trained language, but not the one it was originally trained in. Works quite well for many voices.