[FEAT] Train german Thorsten-Voice model

[thorstenMueller]

Objective

Hi @NeonClary,

i think your existing european language models have a really fair balance between speed (RTF) and quality. Can i train a new model with your code/specs?
I'd like to train a german model based on my Thorsten-Voice dataset (https://zenodo.org/record/7265581) for a good and performant models as yours.

Would this be possible?

Best
Thorsten

Initial Implementation Requirements

• should be compatible with Coqui TTS
• should be useable in Mycroft devices with Neon AI

Other Considerations

This model should be as free as my other models.

[NeonClary]

Hi Thorsten, we're happy to have you build on our work. I'm including Daniel & Bohdan on this email, since this is mainly Bohdan's project, and Daniel is our lead. I mentioned your request in our morning meeting today, and Daniel says he's already "met" you online as well, and is happy to have you working on this. It may be helpful for you to know that Bohdan is in the EU also, in Ukraine, so his time zone will be closer to yours for collaboration. Clary

…

On Sat, Feb 25, 2023 at 12:46 PM Thorsten Müller ***@***.***> wrote: Objective Hi @NeonClary <https://github.com/NeonClary>, i think your existing european language models have a really fair balance between speed (RTF) and quality. Can i train a new model with your code/specs? I'd like to train a german model based on my Thorsten-Voice dataset ( https://zenodo.org/record/7265581) for a good and performant models as yours. Would this be possible? Best Thorsten Initial Implementation Requirements - should be compatible with Coqui TTS - should be useable in Mycroft devices with Neon AI Other Considerations This model should be as free as my other models. — Reply to this email directly, view it on GitHub <#100>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AX4YFARDA2H53ITJH4LJWE3WZJVQ5ANCNFSM6AAAAAAVIBUHAI> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[thorstenMueller]

Hi @NeonClary ,
thanks for your response. As i can't see the real mailadresses of Daniel and Bohdan (just anonymized Github addresses) i'll respond directly here. But if you prefer on communicating by mail please send me your contact data (https://www.thorsten-voice.de/en/contact/) for talking about next steps.
Thanks for your support 😊