Finetuned openai/whisper-small on 3620 Greek training audio samples from mozilla-foundation/common_voice_17_0.

This model was created from the Mozilla.ai Blueprint: speech-to-text-finetune.

Evaluation results on 1701 audio samples of Greek:

Baseline model (before finetuning) on Greek

• Word Error Rate: 46.392
• Loss: 0.902

Finetuned model (after finetuning) on Greek

• Word Error Rate: 45.632
• Loss: 0.869

Finetuned model (after finetuning) on the Greek FLEURS test set (total of 649 samples)

• Word Error Rate (Normalized): 35.66
• Word Error Rate (Orthographic): 38.337
• Character Error Rate (Normalized): 14.041
• Character Error Rate (Orthographic): 13.634
• Loss: 1.3