Intelligent Risk Identification through Speech

Just as the iris of your eye adjusts to let the perfect amount of light reach your retina, our IRIS model skillfully streamlines the flow of information from earnings calls.

At the core of IRIS is a cutting-edge deep neural network that has been trained on a globally sourced dataset developed by professional psychologists specifically for emotion recognition. This training dataset encompasses millions of experimentally controlled expressions collected across dozens of unique cultures and encapsulating millions of hours of human interaction. This rich training background empowers IRIS to accurately distinguish individual speakers and analyze their vocal expressions across hundreds of nuanced dimensions.

Corporate earnings calls are a tightly managed and choreographed environment. However, while the statements delivered by corporate executives on these calls are carefully vetted, the underlying sound structure of the executive's voice is relatively uncontrolled and unconscious. This presents an opportunity for IRIS to pick up on informational “leaks” in which critical insights can be derived from vocal patterns even when the information being delivered was meant to be tightly managed.

IRIS enhances traditional analysis by tapping into the rich data source provided by executive speech. It breaks speech down to its fundamental components to decode the complex web of emotional conveyance embedded beneath the surface-level words. With the raw vocal patterns, IRIS is then able to pinpoint specific moments of abnormal emotional fluctuations and understand the business topics that were being discussed at that time. Importantly, IRIS acknowledges the individuality of vocal expression. It creates a personalized baseline profiles for each executive speaker (derived from previous earnings calls) against which current vocal patterns are compared and abnormal deviation patterns are identified.

IRIS simplifies this complex process by categorizing its analysis into three distinct groups: positivity, negativity, and uncertainty. This simplification aids in quantifying emotional undertones, providing an objective and measurable understanding of the speaker’s emotional state. Moreover, IRIS identifies critical statements and contextualizes them, highlighting their significance in the business narrative.

The sections below further detail each of IRIS's unique features.

Next-Gen Risk Evaluation

IRIS is designed to analyze each statement from the executive team, providing a risk assessment score based on the emotional content of the speech. This score is derived from a combination of factors, including the speaker's vocal tone, pitch, and speed, as well as the overall sentiment of the statement.

Risk Assessment Checklist

Critical Statement Extraction

Our model automates the extraction of key CEO statements from earnings calls, pinpointing crucial insights for investors without having to manual sift through lengthy discussions.

Alternative investment analysis

Start Investing With IRIS Today

Join our global community of investors who are using IRIS to maximize their risk-adjusted returns. IRIS is the only model of its kind and can only be found on EarningsEdge.ai.

  • Only earnings call analysis model of its kind
  • Extensively trained on global dataset
  • Accurate risk detection
  • Powerful visualization tools
  • Automates critical insights
  • CEO baseline profiles