← All examples

ModelEthio-ASR Amharic (600M)2Ethio-ASR Multi (300M)4Ethio-ASR Multi (600M)4Ethio-ASR Multi (94M)4Meta MMS-1B9OmniASR-CTC (1B)5OmniASR-CTC (300M)7OmniASR-CTC (3B)0OmniASR-LLM (1B)0OmniASR-LLM (300M)0OmniASR-LLM (3B)0OpenAI Whisper-small30
Audio qualityClean Studio5Ambient Room Noise2Harsh Environment2Distant / Low-Volume Mic5Wideband Phone (VoLTE)4Randomized Phone Channel2Narrowband 2G + Packet Loss9

What did the model get wrong?

OmniASR-CTC (1B) Distant / Low-Volume Mic   Example 6 of 20  Β·  speaker: Male

🎧 Listen β€” Distant / Low-Volume Mic
Very quiet audio, like the microphone was far away or turned down.
βœ… What was actually said
αˆ˜α‰Έ αŠα‹ α‹¨αˆšαˆ΅α‰°αŠ«αŠ¨αˆˆα‹ ሡራ αˆ΅αˆ‹αˆˆα‰₯ኝ α‰ΆαˆŽ α‰’αˆ†αŠ•αˆαŠ αŒ₯ሩ αŠα‹
πŸ€– What this model heard
αˆ˜α‰ΈαŠ α‹¨αˆšαˆ΅α‰°αŠ«αŠ¨αˆˆα‹ ሡራ αˆ΅αˆ‹αˆˆα‰₯ኝ α‰΅αˆŽ α‰’αˆ†αŠ ልኝ αŒ₯ሩ αŠα‹
Correct
Wrong (different word)
Missed (skipped)
Extra (added)
5
Correct
3
Wrong
1
Missed
1
Extra
56%
Words right
Out of 9 spoken words, this model got 5 right (56%). Errors: 3 words wrong; 1 missed; 1 extra.

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