Synthetic English speech dataset generated with openbmb/VoxCPM2.
TTS fine-tuning
voice-cloning research
synthesis benchmarking
stability, robustness, and post-processing experiments
The audio in this release is synthetic. It is not a corpus of naturally recorded human speech.
train/metadata.csv: public release manifest… See the full description on the dataset page:
https://huggingface.co/datasets/owensong/voxcpm2-synthetic-en-v1.