โ† All examples

ModelEthio-ASR Amharic (600M)8Ethio-ASR Multi (300M)16Ethio-ASR Multi (600M)11Ethio-ASR Multi (94M)10Meta MMS-1B12OmniASR-CTC (1B)7OmniASR-CTC (300M)13OmniASR-CTC (3B)9OmniASR-LLM (1B)5OmniASR-LLM (300M)11OmniASR-LLM (3B)10OpenAI Whisper-small12
Audio qualityClean Studio6Ambient Room Noise7Harsh Environment3Distant / Low-Volume Mic7Wideband Phone (VoLTE)9Randomized Phone Channel10Narrowband 2G + Packet Loss16

What did the model get wrong?

Ethio-ASR Multi (300M) Narrowband 2G + Packet Loss   Example 20 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
แ‰แŒฅแˆฉ แ‹œแˆฎ แ‹˜แŒ แŠ แˆแˆˆแ‰ต แ‹œแˆฎ แŠ แˆแˆตแ‰ต แˆตแ‹ตแˆตแ‰ต แˆฐแ‰ฃแ‰ต แˆตแˆแŠ•แ‰ต แ‹˜แŒ แŠ แ‹œแˆฎ แŠแ‹
๐Ÿค– What this model heard
sid tiua daro bitane a kulati harao a amahinai aa maati sanao ati kaha na mano
Correct
Wrong (different word)
Missed (skipped)
Extra (added)
0
Correct
12
Wrong
0
Missed
4
Extra
0%
Words right
Out of 12 spoken words, this model got 0 right (0%). Errors: 12 words wrong; 4 extra.

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