AI and Data Science Master the art of data science. Join now
This is for the Arabic language. The client has been using Watson Speech to Text for Transcriptions, for the past year.
We would like to know whether the Watson STT API is learning from the audio files being fed into the system for transcription. Does the Watson STT service's AI capability learn while transcribing Arabic speech to text so that it does not make the same mistake again. Ideally it should keep learning from the uploads and not make the same mistake. In reality what we found is that the mistakes persist. This is important because the team reviewing the text output must progressively spend less time reviewing the output and correcting the mistakes.Can someone please help me on the issue above? Any experience in the past. Thanks.
Hey Sana, I Know I am pretty late to reply to this but still wanted to share the solution.
While Watson STT API is designed to learn from audio files for better accuracy in Arabic transcription, the system might not eliminate all mistakes as expected. Continuous improvements are aimed for, but manual review for correction might still be necessary. Seeking support or adjusting settings could help enhance the learning process for improved accuracy over time.
If the problem still occurs, then try using other transcription tools available online.
Hope this helps!