← All examples

ModelEthio-ASR Amharic (600M)8Ethio-ASR Multi (300M)18Ethio-ASR Multi (600M)9Ethio-ASR Multi (94M)9Meta MMS-1B9OmniASR-CTC (1B)9OmniASR-CTC (300M)8OmniASR-CTC (3B)4OmniASR-LLM (1B)8OmniASR-LLM (300M)8OmniASR-LLM (3B)9OpenAI Whisper-small9
Audio qualityClean Studio0Ambient Room Noise2Harsh Environment2Distant / Low-Volume Mic0Wideband Phone (VoLTE)2Randomized Phone Channel2Narrowband 2G + Packet Loss8

What did the model get wrong?

OmniASR-LLM (1B) Narrowband 2G + Packet Loss   Example 7 of 20  Β·  speaker: Female

🎧 Listen β€” Narrowband 2G + Packet Loss
Heavy background noise plus phone-call compression β€” worst-case stress test.
βœ… What was actually said
α‹¨αŠ’αŠ•α‰°αˆ­αŠ”α‰΅ αŒ₯α‰…αˆ α‹ˆα‹­αˆ α“αŠ¬αŒ… αˆ˜αŒα‹›α‰΅ αŠ₯α‰½αˆ‹αˆˆαˆ α‹‹αŒ‹α‹ αˆ΅αŠ•α‰΅ αŠα‹
πŸ€– What this model heard
α‹­αˆ… α‹¨αˆα‰΅α‰€αˆ የαŠ₯ሡከ αŠ αŠ«α‰£α‰’α‹ˆα‰½ αŠ“α‰Έα‹ α‹¨αˆšαŒˆαŠ˜α‹ የውቀው αŠ₯αŒ… αŠα‹
Correct
Wrong (different word)
Missed (skipped)
Extra (added)
1
Correct
8
Wrong
0
Missed
0
Extra
11%
Words right
Out of 9 spoken words, this model got 1 right (11%). Errors: 8 words wrong.

Try switching Audio quality β€” see the same model degrade as the audio gets noisier.