Come join us on June 2nd as we meet Thorsten Müller to talk about the experience of publishing an open neural text-to-speech dataset in their own voice.

Conversational AI redefines modes of human-machine interaction in many scenarioes, and high-quality, expressive and efficient speech synthesis is a rapidly expanding research area in this transformation. However, it is not always easy to find good open datasets for languages other than English as planning, recording, pre- and post-processing a TTS dataset may require several considerations.

Thorsten  is a truely open, high-quality TTS dataset in German. It was recorded and made public by Thorsten Müller under the terms of Creative Commons Zero V1 Universal (CC0), which is used to opt out of copyright entirely and ensure that the work has the widest reach. This is a great contribution to the community and has attracted the attention of many researchers and engineers. You can find several pretrained models with Thorsten dataset, even on TensorFlow Hub.

In this event, Yusuf Sarıgöz will host Thorsten Müller to discuss about their motivation and experience of publishing a TTS dataset in their own voice from idea to publication and to followup datasets such as "Thorsten Emotional", a recent extension to the dataset for emotional TTS. We strongly believe that this talk will be inspirational and fruitful with a lot of learning opportunities.

It will be live broadcasted on our YouTube channel, so RSVP to the event to get updates.