โ† All examples

ModelEthio-ASR Amharic (600M)6Ethio-ASR Multi (300M)15Ethio-ASR Multi (600M)7Ethio-ASR Multi (94M)7Meta MMS-1B14OmniASR-CTC (1B)8OmniASR-CTC (300M)9OmniASR-CTC (3B)7OmniASR-LLM (1B)6OmniASR-LLM (300M)6OmniASR-LLM (3B)7OpenAI Whisper-small7
Audio qualityClean Studio3Ambient Room Noise3Harsh Environment1Distant / Low-Volume Mic3Wideband Phone (VoLTE)4Randomized Phone Channel1Narrowband 2G + Packet Loss7

What did the model get wrong?

OmniASR-CTC (3B) Narrowband 2G + Packet Loss   Example 10 of 20  ยท  speaker: Male

๐ŸŽง Listen โ€” Narrowband 2G + Packet Loss
Heavy background noise plus phone-call compression โ€” worst-case stress test.
โœ… What was actually said
แˆตแˆˆ แŒฅแ‰…แˆŽแ‰น แ‹แˆญแ‹แˆญ แˆ˜แˆจแŒƒ แ‹จแ‰ต แАแ‹ แ‹จแˆ›แŒˆแŠ˜แ‹
๐Ÿค– What this model heard
al ai no magao
Correct
Wrong (different word)
Missed (skipped)
Extra (added)
0
Correct
4
Wrong
3
Missed
0
Extra
0%
Words right
Out of 7 spoken words, this model got 0 right (0%). Errors: 4 words wrong; 3 missed.

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