Source: Baik, Kim, Kim & Yoon (2025) — “Vocal Delivery Quality in Earnings Conference Calls”
I like finance papers that accidentally become papers about human behaviour. This one does that. It looks like a study about data, markets, or disclosure, but underneath it is really about how people decide what to believe.
One of the most interesting aspects of earnings call research is the growing attention to how information is communicated, not just what is communicated. This paper looks at something most analysts would struggle to articulate: the acoustic quality of how CEOs and CFOs actually sound when they speak. The researchers introduce a measure of vocal delivery quality using deep learning applied to audio recordings of earnings calls. What they’re capturing isn’t accent or pleasantness — it’s acoustic comprehensibility: how easy the audio is to process for an average listener, based on properties like clarity, pace, and consistency of delivery. The finding that struck me: vocal delivery quality deteriorates when executives are delivering bad news. Specifically, quality goes down when earnings are declining or when the narrative content is negative. It also drops when earnings are positive but the increase is transitory — a kind of temporary beat that doesn’t persist. In other words, executives communicate differently when the news is bad or when positive results are misleading, and this shows up in measurable acoustic properties, not just in what words they choose. There are two ways to interpret this.
In plain English, that is why the result matters beyond the chart. It changes where people should look, what they should question, and which comfortable assumption probably needs to be retired.
My takeaway: AI and data do not remove judgment from finance. They move judgment to the part where people decide what the signal means, whether it is trustworthy, and whether they are brave enough to act on it.