At Speechmatics, the company is continually working on improving its Autonomous Speech Recognition (ASR) capabilities by incorporating more nuanced understanding of emotions in speech-to-text. The team tested their ASR technology on the popular TV show Friends, analyzing 1400 dialogues and 13000 utterances from various characters, including Ross, Rachel, Monica, Chandler, Phoebe, and Joey. The results showed that fear is the most challenging emotion for ASR to recognize, with an average accuracy of 78%, compared to neutral emotions which reached 85% accuracy. However, even among other emotional states, minor differences in accuracy were found, likely due to chance. The team also discovered that certain characters, like Phoebe and Joey, have unique voices that significantly impact their ASR accuracy, especially when it comes to positive surprise and fear. To improve speech-to-text technology, Speechmatics aims to incorporate more diversity into its systems, recognizing the importance of emotions in shaping individual differences in voice and everyday life.