Challenge Tracks

Track 1 - Streaming TTS

In this track, the participants should build low latency TTS systems, using challenge data along with open-source data or finetuned on open-source pre-trained models.

Track 2 - Neural codec TTS

In this track, the participants should build TTS systems with neural codecs, with codecs representing at least 3 attributes like speaker identity, content, pitch, energy etc. The TTS model and neural codec model can be a trained open-source model or can be trained on challenge data + open-source data.

Track 3 - Streaming TTS with neural codecs

In this track, the participants should build low-latency TTS systems with neural codecs. The neural codec model can be a trained open-source model or can be trained on challenge data + open-source data. The TTS can be trained using challenge data along with open-source data or finetuned on open-source pre-trained models.