Ken-Z
/

latin_whisper-small

Automatic Speech Recognition

Model card Files Files and versions

Whisper Small Latin - Ken

This model is a fine-tuned version of openai/whisper-small on the Latin language. It achieves a 20 CER (with punctuations) with the 67 hours of audio from Vox Classica

Try this model directly in your browser: ASR Demo

Downloads last month: 376

Safetensors

Model size

0.2B params

Tensor type

F32

·

Model tree for Ken-Z/latin_whisper-small

Base model

microsoft/speecht5_tts

Finetuned

(1308)

this model

Dataset used to train Ken-Z/latin_whisper-small

Spaces using Ken-Z/latin_whisper-small 2