talromur_3_with_prompts is a prompt-labelled corpus that can be used for fine-tuning models, such as ParlerTTS.The corpus consists of approximately 15,000 utterances, spoken by 7 named speakers in 6 different emotions (see more info here).
The dataset is an expanded version of Talromur-3: an Icelandic emotional speech corpus.We have added natural-language descriptions of utterance-level pitch, speech monotony, speech quality, reverberation, speaking rate, and emotional… See the full description on the dataset page:
https://huggingface.co/datasets/atlithor/talromur3_with_prompts.
