When cutting out a snippet from the original mp3 file, it recognizes it perfectly, confidence levels usually not below 500. But when I try to have it recognize by mic, or recognize a file recorded by mic, it recognizes it totally inaccurately with confidence levels never above 10. I am on Ubuntu 18.04 and am using pavucontrol to record audio from the monitor of system audio. The recorded clips I used were 10 seconds long.
When cutting out a snippet from the original mp3 file, it recognizes it perfectly, confidence levels usually not below 500. But when I try to have it recognize by mic, or recognize a file recorded by mic, it recognizes it totally inaccurately with confidence levels never above 10. I am on Ubuntu 18.04 and am using pavucontrol to record audio from the monitor of system audio. The recorded clips I used were 10 seconds long.