US 12,334,077 B1
Robust methods for automated audio signal to text signal processing systems
Gaurav Iyer, Delhi (IN); Kaushal Agarwal, Pune (IN); Shuvadib Paul, Kolkata (IN); Sandeep Bhutani, Kurukshetra (IN); Ankush Jain, Delhi (IN); and Abdulrehman Sayyad, Pune (IN)
Assigned to ExlService Holdings, Inc., New York, NY (US)
Filed by ExlService Holdings, Inc., New York, NY (US)
Filed on Nov. 26, 2024, as Appl. No. 18/961,316.
Int. Cl. G10L 15/18 (2013.01); G06F 9/451 (2018.01); G10L 15/01 (2013.01); G10L 15/22 (2006.01); G10L 15/26 (2006.01); H04L 67/306 (2022.01)
CPC G10L 15/26 (2013.01) [G06F 9/451 (2018.02); G10L 15/01 (2013.01); H04L 67/306 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method performed by a signal processing system, the method comprising:
receiving audio signal data corresponding to a recorded interaction involving at least one authorized provider of an enterprise service and at least one subscribing user of the enterprise service,
wherein the at least one subscribing user of the enterprise service is associated with a user service profile;
converting, using a machine learning model, the audio signal data into a transcript comprising alphanumeric components;
prompting a generative machine learning model to generate a response that maps at least one alphanumeric component of the converted transcript to a target signal domain group,
wherein configuration of the prompt for the generative machine learning model comprises a set of identifiable user features from the user service profile, and
wherein the target signal domain group comprises:
at least one signal extraction rule for determining target alphanumeric elements within alphanumeric components of the converted transcript, and
at least one signal processing criterion for evaluating performance scores associated with the at least one authorized provider;
prompting the generative machine learning model to generate a response that includes a set of alphanumeric elements from the at least one alphanumeric component, the set of alphanumeric elements satisfying the at least one signal extraction rule of the target signal domain group;
predicting, using the determined set of alphanumeric elements, a set of sentiment attributes corresponding to the at least one alphanumeric component of the converted transcript;
estimating, via comparison of the set of alphanumeric elements and the set of sentiment attributes to the at least one signal processing criterion, a signal performance score of the at least one authorized provider; and
displaying, at a user interface, a custom interface component enabling an authorized user to modify a set of configurable interface elements,
wherein the set of configurable interface elements comprises (1) a first user configurable element associated with the signal performance score and (2) a second user configurable element associated with the set of alphanumeric elements.