luel/emofi-tts
Text-to-Speech • Updated
You need to agree to share your Hugging Face information to access this dataset
This repository is publicly accessible, but you have to accept the conditions to access its files and content.
This package is derived from the MEAD dataset and is distributed under the MEAD research license. By requesting access you confirm that you will use it for non-commercial research only and that you have agreed to the MEAD Terms of Use at https://github.com/uniBruce/Mead/blob/master/Terms%20of%20Use_%20Mead.pdf.
Log in or Sign Up to review the conditions and access this dataset content.
EmoFI-MEAD
Preprocessed MEAD package used to train and evaluate EmoFI-TTS (arXiv:2610.30477). Code and instructions: github.com/luelhagos/Emofi. Weights: luel/emofi-tts.
Files
| File | Contents | Unzip to (repo root) |
|---|---|---|
MEAD_features.zip |
Mel spectrograms, pitch, energy, durations, cached face (768-d) and ECAPA speaker (192-d) embeddings, train/val/test manifests, speaker and phoneme tables | data/MEAD/ |
MEAD_test_wavs.zip |
The real test-set utterances at 22.05 kHz (real-audio ceilings, reference clips, speaker-similarity targets) | data/MEAD_processed/ |
MEAD_corpus.zip |
Per-utterance wavs, one frontal face frame per utterance, and transcripts for the full subset; needed only to rebuild the features from scratch | data/MEAD_corpus/ |
hf auth login
hf download luel/emofi-mead --repo-type dataset --local-dir data
Coverage
English, 46 of the 48 MEAD speakers (two excluded for incomplete emotion-level coverage), six emotions (neutral, angry, happy, sad, surprised, fear) at three intensity levels, split speaker-disjointly 36 / 4 / 6 for train / val / test. Disgust and contempt are excluded. Transcripts were content-matched to MEAD's canonical sentence list and verified with Whisper-large; alignments come from the Montreal Forced Aligner.
License
MEAD is released for non-commercial research. This package is a derivative and carries the same terms; redistribution outside those terms is not permitted.
Citation
Please cite both MEAD and EmoFI-TTS:
@inproceedings{wang2020mead,
title = {MEAD: A Large-scale Audio-visual Dataset for Emotional Talking-face Generation},
author = {Wang, Kaisiyuan and Wu, Qianyi and Song, Linsen and Yang, Zhuoqian and Wu, Wayne and Qian, Chen and He, Ran and Qiao, Yu and Loy, Chen Change},
booktitle = {ECCV},
year = {2020}
}
@article{jeon2026emofi,
title = {EmoFI-TTS: Face-Conditioned Expressive Voice Synthesis with Intensity Modulation},
author = {Jeon, Yejin and Beyene, Luel H. and Maltais, Marie and Kim, Hyounghun and Adelani, David I.},
journal = {arXiv preprint arXiv:2610.30477},
year = {2026},
url = {https://arxiv.org/abs/2610.30477}
}
- Downloads last month
- 29
Models trained or fine-tuned on luel/emofi-mead