DatasetLwaziANV

← All examples Model summary →

Modelfacebook/omniASR-CTC-1BOpen8facebook/omniASR-CTC-300MOpen7facebook/omniASR-CTC-3BOpen9facebook/omniASR-LLM-1BOpen8facebook/omniASR-LLM-300MOpen7facebook/omniASR-LLM-3BOpen12facebook/omniASR-LLM-7BOpen3dsfsi/anv-whisper-large-v3-turbo-anv-zulOpen10dsfsi/anv-whisper-small-anv-zulu-first-batchOpen13openai/whisper-large-v3Open38openai/whisper-smallOpen30sitwala/whisper-large-anv-sotOpen47sitwala/whisper-large-v3-anv-sotOpen55sitwala/whisper-large-v3-turbo-anv-zul-150hOpen9sitwala/whisper-large-v3-turbo-anv-zul-250hOpen10sitwala/whisper-large-v3-turbo-anv-zul-50hOpen11

What did the model get wrong?

sitwala/whisper-large-v3-turbo-anv-zul-250hOpen   Example 38 of 100  ·  sample index 37

✅ What was actually said (isiZulu reference)
AmathempeliamaGrekinamathempeliamaRomaayebhekenenetshelemabulaelimhlophe,futhikusukelangekhulule-18,ngokufikakwezakhiwoze-neoclassical,umbalaomhlophewabaumbalaovamekakhuluwamasontoamasha,ama-capitol,nezinyeizakhiwozikahulumeni,ikakhulukazie-UnitedStates.
🤖 What this model heard
amathempeliamagreeknamathempeliamaromaayebhekenenetshelemabulaelimhlophefuthikusukelangelikhulule18ngokufikakwezakhiwozeneoclassicalumbalaomhlophewabaumbalaovamekakhuluwamasontoamashaamakhephitholinezinyeizakhiwozikahulumeni
Correct
Wrong (different word)
Missed (skipped)
Extra (added)
22
Correct
5
Wrong
3
Missed
2
Extra
73%
Words right
0.433
WER
0.166
CER
33.72s
Duration
0.1s
Latency