Datasets:

---

amphion

/

Emilia-NV

like 33

Follow

Amphion 397

I read the NVSpeech paper, and from what I understand, this dataset is a sub-partition of Emilia, to which a specialized non-verbal ASR model has been applied.
However, I would like to train a non-commercial TTS on it, as well as on a partition of Emilia Chinese. How can I distinguish between the Emilia-zh and Emilia-nv data blocks so that I don't have duplicates in my dataset?

Thank you to the Amphion team for making this dataset public 👍