We just wanted to automate our Python intern hiring process by using Amazon transcribe ML tool.
At first, we thought it’s gonna be ground-breaking and we could do most of our hiring process by getting text and funnel to our in-house tech-scorer algorithm.
However, when we recorded the audio and funneled to Amazon Transcribe. We got really shocked when we got the result.
Here is the original audio:
This is what we have got from Amazon Transcribe.
Speaker 0: So what if I turn more?Speaker 1: Create a brighter mornings of file containing the fighter coldSpeaker 1: so the model can defend functions, classes and variables.Speaker 1: So, after better be candid on somebody.Speaker 0: Okay, Well, what is the difference between a list on to pulls and pipe bombs?Speaker 1: Oh, this little nipples are immutable. Where lives are miserable. In ***,Speaker 1: we can createSpeaker 1: people's in dictionary on God use list US keys individually.Speaker 1: Um, there are 33 available pupils methods in baton. And there are 46Speaker 1: the methods available inducedSpeaker 0: What is self in fight on.Speaker 1: So basically refers to toe instance off a class where it buying certain foods and argumentsSpeaker 1: can we can use silver as a parameter on. But which of us, to the instance off plus on become access all its variable somebody boots.Speaker 0: Okay. Thank you.
Obviously, you could see there is no match at all. Finally, we have concluded that Amazon Transcribe is not the right way to hire our Python Interns.
We are working on our in-house speech2text algorithms with the help of open-source code. We will be publishing the details soon.
Note: If you find that we are missing any input parameters to Amazon Transcribe, please let us know. We are happy to learn from our mistakes.
