Safetensors
whisper

Malaysian Finetune Whisper Large V3 Turbo Non-Verbal

Extension of mesolitica/malaysian-whisper-large-v3-turbo-v3 trained on mesolitica/Speech-Nonverbal-Whisper

Improvement

  1. Better on filler predictions, introduced <|transcribenonverbal|> token, a new task!

Non-verbal part of the dataset but currently the model is underperformed to predict non-verbal such as laughing.

Downloads last month
0
Safetensors
Model size
809M params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for mesolitica/malaysian-whisper-large-v3-turbo-v3-nonverbal

Datasets used to train mesolitica/malaysian-whisper-large-v3-turbo-v3-nonverbal