Four models,
one honest score.
What is actually running while you talk: the methods and machine-learning models behind every rep, and what each one is listening for.
Every word, on your device.
Your answers are transcribed by a voice recognition model running on your own device. Working from the transcript, each answer is assessed on its own rather than as one long block of talking.
That is how we can tell whether your answer to “what salary are you looking for” was a sentence or a speech, and point you at the words that would have landed better.
How fast, and how soon.
We measure the speed you talk at, and how long it takes the algorithm to catch your first word after the question lands.
In an interview you want to come across as someone the panel can settle in with, speaking at a pace they can follow, and giving context before the detail. Words per minute and time-to-first-word are the two numbers that move that most.
The attitude behind the words.
A machine-learning model reads the tone of the transcript to work out the attitude a candidate is bringing to each answer.
Staying positive is a good sign you would be an easy person to have on the team. But some answers need a different register, and we help you pick the words that carry the weight the moment calls for.
The flat delivery you cannot hear.
Pitch analysis follows your vocal frequency in real time and tracks every change in the pitch of your voice.
Plenty of people speak flat without ever knowing it, because you cannot hear your own intonation while you are using it. Seeing the trace is usually all it takes to fix.
Do the first rep tonight.
Ten minutes, no card, one honest score, with the detail behind every point of it.
Start free repHow coaching works →