โ† All examples

ModelEthio-ASR Amharic (600M)2Ethio-ASR Multi (300M)1Ethio-ASR Multi (600M)1Ethio-ASR Multi (94M)1Meta MMS-1B8OmniASR-CTC (1B)2OmniASR-CTC (300M)3OmniASR-CTC (3B)0OmniASR-LLM (1B)2OmniASR-LLM (300M)0OmniASR-LLM (3B)0OpenAI Whisper-small15
Audio qualityClean Studio8Ambient Room Noise8Harsh Environment8Distant / Low-Volume Mic8Wideband Phone (VoLTE)8Randomized Phone Channel8Narrowband 2G + Packet Loss8

What did the model get wrong?

Meta MMS-1B Clean Studio   Example 11 of 20  ยท  speaker: Female

๐ŸŽง Listen โ€” Clean Studio
Clear, studio-quality recording. Best-case audio.
โœ… What was actually said
แŠฅแˆฝ แ‰ แŒฃแˆ แŠจแแ‰ฐแŠ›แ‹ แแŒฅแŠแ‰ต แŠฅแˆตแŠจ แˆตแŠ•แ‰ต แ‹ญแ‹ฐแˆญแˆณแˆ แ‹‹แŒ‹แ‹แˆต
๐Ÿค– What this model heard
แˆญ แ‰ แˆˆ แˆญแ‰  แŒˆ แŠ“แŠ
Correct
Wrong (different word)
Missed (skipped)
Extra (added)
0
Correct
5
Wrong
3
Missed
0
Extra
0%
Words right
Out of 8 spoken words, this model got 0 right (0%). Errors: 5 words wrong; 3 missed.

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