DatasetLwaziANV

← All examples Model summary →

Modelfacebook/omniASR-CTC-1BOpen18facebook/omniASR-CTC-300MOpen21facebook/omniASR-CTC-3BOpen19facebook/omniASR-LLM-1BOpen16facebook/omniASR-LLM-300MOpen22facebook/omniASR-LLM-3BOpen15facebook/omniASR-LLM-7BOpen12dsfsi/anv-whisper-large-v3-turbo-anv-zulOpen19dsfsi/anv-whisper-small-anv-zulu-first-batchOpen26openai/whisper-large-v3Open42openai/whisper-smallOpen148sitwala/whisper-large-anv-sotOpen54sitwala/whisper-large-v3-anv-sotOpen55sitwala/whisper-large-v3-turbo-anv-zul-150hOpen23sitwala/whisper-large-v3-turbo-anv-zul-250hOpen23sitwala/whisper-large-v3-turbo-anv-zul-50hOpen27

What did the model get wrong?

sitwala/whisper-large-v3-turbo-anv-zul-150hOpen   Example 41 of 100  ·  sample index 40

✅ What was actually said (isiZulu reference)
Ngoqiniso,i-Kode9,uSteveGoodman,ongumqeqeshiye-elektronikhi,uhambaphambiliekuboniseniisithakazeloesiphakemekwi-labelye-musicye-gqom,i-Hyperdub.Ngaphezukwalokho,amathrekhiase-Gqomayatholakalakwi-EP"Touch"KGandi-ScratchaDVA,kanyeno-andBagaManka-Scratchclart.
🤖 What this model heard
ngoqinisoicode9ustevegoodmanongumqeqeshiyeelektronikhiuhambaphambiliekuboniseniisithakazeloesiphakemekwilabelyemusicyegqomihyperdubngaphezukwalokhoamatrackiasegqomayatholakalakwieptouchkgandiscratcherdvakanye
Correct
Wrong (different word)
Missed (skipped)
Extra (added)
19
Correct
8
Wrong
5
Missed
10
Extra
59%
Words right
0.906
WER
0.220
CER
35.62s
Duration
0.1s
Latency