File size: 410 Bytes
7eb4e5d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 |
---
language:
- sr
tags:
- audio
- automatic-speech-recognition
license: mit
datasets:
- juzne-vesti-srpski
- google/fleurs
- mozilla-foundation/common_voice_16_1
- espnet/yodas
---
# Model
Fine-tune of openAI's [whisper-medium](https://huggingface.co/openai/whisper-medium) on the multiple datasets.
It achieves the following results on the evaluation set:
- Loss: 0.1358
- Wer Ortho: 0.1814
- Wer: 0.0901 |