Most people who try automated transcription services are disappointed by the “accuracy.”
But the single biggest factor in transcription accuracy isn’t the transcription itself – most services can accurately transcribe clean audio at 95%+.
So what matters more than raw transcription accuracy?
the ability of the service to accurately understand who is speaking when
Most services use open-source software to “guess” who is speaking by analyzing differences in pitch and tone. Sounds cool, right? Unfortunately it doesn’t work. Typical problems include:
- their systems can’t tell the difference between similar sounding speakers
- they often “guess” more speakers are on the call than actually are
- they aren’t able do speaker-level analysis to understand if one caller was angry or excited – this can only be done at the conversation-level
Fortunately, there is a better way. We’ll share in a future post how Jog developed an entirely new approach to deliver, for the first time, 100% accurate speaker-splitting.
For now, take it from customers – Jog is the most accurate transcription platform available: