Emova-ollm/emova-asr-tts-eval

EMOVA-ASR-TTS-Eval 🤗 EMOVA-Models | 🤗 EMOVA-Datasets | 🤗 EMOVA-Demo 📄 Paper | 🌐 Project-Page | 💻 Github | 💻 EMOVA-Speech-Tokenizer-Github Overview EMOVA-ASR-TTS-Eval is a dataset designed for evaluating the ASR and TTS performance of Omni-modal LLMs. It is derived from the test-clean set of the LibriSpeech dataset. This dataset is part of the EMOVA-Datasets collection. We extract the speech units using the EMOVA Speech Tokenizer. Structure… See the full description on the dataset page: https://huggingface.co/datasets/Emova-ollm/emova-asr-tts-eval.