Paralinguistic Alpha: Predictive Power of Vocal Bio Metrics

---

Management routinely conveys previously material non-public information to the market during earnings calls, and these disclosures move stock prices. Although communication in this setting is inherently multi-modal, market participants overwhelmingly focus on the words spoken, treating transcripts as sufficient representations of managerial intent and conviction. This paper demonstrates that the voice itself carries a distinct and economically meaningful information channel. Using paralinguistic acoustic features extracted from CEO speech during the Q&A segments of earnings calls, we show that vocal delivery encodes managerial confidence, stress, and uncertainty in ways that are orthogonal to textual content. Employing an event-study framework, we find that these acoustic signals predict short- and medium-horizon excess returns, indicating that markets do not fully incorporate the information conveyed by voice at the time of disclosure. The results establish the voice channel as a material, under-recognized component of corporate communication and a source of alpha for systematic investors, independent of language-based analysis 

 Paralinguistic features, such as assertiveness, arousal, and nervousness, contain significant economic information. Our findings demonstrate that even when language is devoid of strong text sentiment, the acoustic properties of executive speech can predict post-earnings announcement drift 

This research brief examines whether managerial vocal delivery improves the economic interpretation of earnings-call text sentiment. Building on evidence that paralinguistic features of executive speech contain economically relevant information, we analyze 41,395 Russell 3000 earnings-call observations from July 15, 2020 through September 30, 2025, covering 2,862 unique tickers. Rather than asking whether voice predicts returns when text is neutral, this study examines whether vocal delivery helps investors interpret explicitly positive and negative textual sentiment. We sort earnings-call observations by NLP sentiment and management-calibrated vocal measures, then evaluate event-sample-relative excess returns over 10-, 20-, and 30-trading-day horizons. The findings show that vocal delivery adds incremental information to text sentiment, with the strongest evidence on the downside. In the lowest NLP sentiment quintile, negative text delivered with high Weak Constructive Delivery underperforms negative text delivered with low Weak Constructive Delivery by approximately 103 basis points over 20 trading days, with an event-date clustered t-statistic of -3.13. Low Balanced Delivery produces similar downside separation. Positive text also benefits from a voice layer: in the highest NLP sentiment quintile, positive text paired with high vocal Valence generates an additional 39 basis points over 10 trading days relative to positive text alone, with a clustered t-statistic of 2.03. The results suggest that vocal delivery is not a substitute for NLP sentiment, but an incremental conditioning layer that helps distinguish routine positive or negative language from language delivered with confirming vocal tone, controlled delivery, or weaker constructive vocal support. 

Asset managers increasingly explore the idea of extracting confidence, stress, or tone from earnings-call audio. Many already license the raw inputs: audio recordings and transcripts, and assume that with modern ML tools, building proprietary behavioral signals is a straightforward extension. It is not.

Access to audio and text is necessary, but it is nowhere near sufficient.

Across firms, internal build attempts repeatedly fail for the same fundamental reasons.