โ† All examples

ModelEthio-ASR Amharic (600M)4Ethio-ASR Multi (300M)18Ethio-ASR Multi (600M)10Ethio-ASR Multi (94M)10Meta MMS-1B10OmniASR-CTC (1B)10OmniASR-CTC (300M)10OmniASR-CTC (3B)10OmniASR-LLM (1B)8OmniASR-LLM (300M)8OmniASR-LLM (3B)9OpenAI Whisper-small15
Audio qualityClean Studio2Ambient Room Noise0Harsh Environment2Distant / Low-Volume Mic1Wideband Phone (VoLTE)1Randomized Phone Channel5Narrowband 2G + Packet Loss9

What did the model get wrong?

OmniASR-LLM (3B) Narrowband 2G + Packet Loss   Example 12 of 20  ยท  speaker: Female

๐ŸŽง Listen โ€” Narrowband 2G + Packet Loss
Heavy background noise plus phone-call compression โ€” worst-case stress test.
โœ… What was actually said
แŠ แˆแŠ• แ‹ซแˆˆแˆแ‰ต แŠ แ‹ณแˆ› แŠแ‹ แ‹ˆแ‹ฐ แŒ…แ‰กแ‰ฒ แˆˆแˆตแˆซ แˆแˆ„แ‹ต แˆตแˆˆแˆ†แŠ แŠแ‹
๐Ÿค– What this model heard
แŠ แ‹ญแŠแ‰ต แŠแ‹ แŠ แ‹ณแˆ›แŠแ‹ แ‹ˆแ‹ฐ แ‰ตแŠฉแˆต แ‹ญแŒˆแŠ›แˆ แˆ‚แ‹ฐแ‰ต แŠแ‹ แŠฅแŠ“
Correct
Wrong (different word)
Missed (skipped)
Extra (added)
1
Correct
8
Wrong
1
Missed
0
Extra
10%
Words right
Out of 10 spoken words, this model got 1 right (10%). Errors: 8 words wrong; 1 missed.

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