AI-controlled voice optimization CTI

Abstract

We provide a CTI (Computer Telephony Integration) system that combines AI-powered voice conversion control and customer information analysis to assist with making calls or to perform AI-powered automated calling and response. [Solution] In a cloud-based or on-premise CTI system, a customer information acquisition mechanism acquires industry, company size, job title, past call results, etc., and an analysis engine analyzes performance indicators for each voice conversion mode and recommends the optimal mode. The voice conversion engine can convert and output operator voice or AI-generated voice in real time and conducts calls according to the recommended mode. Furthermore, a learning mechanism analyzes the call results, and the analysis engine continuously improves the accuracy of voice mode recommendations, thereby achieving uniformity in call quality and improved call outcomes through optimal voice quality and speaking style according to customer attributes.

Description

Technical Field

[0001] The present invention relates to voice communication technology and call center systems, and particularly to a CTI (Computer Telephony Integration) system that combines voice conversion control by artificial intelligence and customer information analysis to provide call support or perform automatic call connection / automatic response by artificial intelligence. The present invention is applicable to any of cloud-based systems, on-premises systems, or hybrid configurations.

Background Art

[0002] In new business development sales for corporations and call center operations, it is important to give a good impression to the receptionists and decision-makers of customer companies in a short time. However, in conventional CTI systems, the influence of the voice quality and speaking style of the call connector on the call result cannot be fully considered, and the passing rate and negotiation conversion rate depend on the individual ability of the call connector. In recent years, although automatic call connection (voice bots) and automatic response systems by artificial intelligence have become popular, in these artificial intelligence voices, a mechanism for dynamically controlling the optimal voice quality and speaking style according to the customer industry and attributes has not been fully established. Furthermore, in any environment of cloud-based or on-premises, the operation of integrating the voice conversion function and customer information analysis function to improve call results has not been fully realized. Description of the Invention

[0003] Problems to be Solved by the Invention: In conventional CTI systems, it is impossible to automatically select the optimal voice quality and speaking style according to customer attributes and past call results, and the call connection results depend on the individual ability of the operator or the fixed settings of artificial intelligence voices. An object of the present invention is to provide a CTI system in which artificial intelligence recommends an optimal voice conversion mode based on information such as the customer's industry and business type information and past call results in order to solve these problems. In the system of the present invention, when a human operator makes a call, voice conversion is performed in real time based on the recommended mode, and when an artificial intelligence voice agent makes an automated call or provides an automated response, the generated voice can be output using the recommended mode. This will simultaneously improve call outcomes, reduce reliance on the individual skills of callers, and optimize AI voice. Means for solving the problem: The CTI system of the present invention is characterized by including the following components. (1)Customer information acquisition mechanism Retrieve customer characteristics such as industry, company size, job title, past call results, and response history from your CRM or customer database (cloud-based / on-premise compatible). (2) Analysis engine (including artificial intelligence) Based on customer characteristics and call / response history acquired by the customer information acquisition mechanism, performance indicators (pass rate, conversion rate, response rate, etc.) are calculated for each voice conversion mode. Furthermore, the system can learn and recommend the optimal voice conversion mode for each customer industry or attribute. (3) Voice mode database It stores multiple voice feature templates (e.g., Calm Male, Clear Female, Smart Female, Dynamic Male, etc.) and defines voice parameters (timbre, speech rate, tone, emotional expression, etc.) for each mode. (4) Voice conversion engine (including TTS function) Based on recommendations from the analysis engine, the input audio is converted or generated in real time and output as a call voice or response voice. The input audio supports operator voice or AI-generated voice (TTS). (5) Dashboard UI The recommended voice mode is displayed on the screen for callers or administrators. Users can either use the recommended mode or manually select a different mode. When the AI ​​agent automatically makes calls and responds, the recommended mode is automatically applied. (6) Learning mechanism The system sends call or response data to an analysis engine to continuously update voice mode effects based on customer industry and attributes. This allows artificial intelligence to optimize the effectiveness of voice conversion modes. Examples of embodiments of the present invention: The following describes some representative embodiments, but these are merely examples that embody the technical concept of the present invention, and other configurations are also included within the scope of the present invention. Embodiment Example 1: Operator-assisted power supply mode (1)Customer information acquisition mechanism The operator terminal retrieves the call target list from the CRM system or list management module. The customer information acquisition mechanism collects industry, company size, job title, past call results, response history, etc., and sends it to the analysis engine. (2) Analysis engine (including artificial intelligence) Based on customer characteristics and call / response history, the system calculates performance indicators (pass rate, response rate, conversion rate, etc.) for multiple templates registered in the voice mode database and recommends the optimal voice conversion mode. (3) Dashboard UI The recommended mode is displayed. The operator can either accept it or manually select a different mode. (4) Voice conversion engine Based on the selected mode, the operator's actual voice is converted and output in real time. Conversion parameters include voice quality, speaking speed, tone, and intelligibility. (5) Learning mechanism The call results are sent to an analysis engine and analyzed by a learning mechanism. The results are used to improve the accuracy of recommendations for future calls. Embodiment Example 2: Automatic Call Mode by Artificial Intelligence Agent (1)Customer information acquisition mechanism The system automatically retrieves a registered call list and collects information such as the industry, company size, job title, past call results, and response history for each target customer. (2) Analysis engine (including artificial intelligence) Based on the acquired data, the system analyzes the effectiveness of multiple templates in the voice mode database and automatically selects the optimal voice conversion mode. (3) Voice conversion engine (including TTS function) Based on the selected mode, operator voices and AI-generated voices (TTS) are converted and generated in real time. The AI ​​agent uses this voice to automatically speak and conduct conversations. (4) Dashboard UI Displays call status and recommended mode application status for administrators. (5) Learning mechanism The system analyzes and stores call results and response content (response rate, conversation duration, keyword extraction, etc.) and uses this information to update voice modes and conversion parameters. (6) Voice Mode Database Based on learning results, the system maintains and updates optimized voice templates tailored to customer attributes and industries. Effects of the invention: According to the present invention, the following effects can be achieved. (1) Calls and responses are made with the optimal voice quality according to customer attributes, improving call outcomes such as pass rate, answer rate, and conversion rate. (2) Through learning, the accuracy of recommendations is continuously improved, making it possible to configure a self-improving CTI system. (3) The dashboard allows for easy selection of voice modes, reducing differences in individual skill levels and standardizing call quality. (4) Even with automated calling and responses by artificial intelligence agents, natural conversation and high response effectiveness are achieved simultaneously. (5) It is easy to integrate with cloud-based and on-premise CTI / CRM systems, and can also be used for marketing analytics. (6) The artificial intelligence model and voice parameters can be updated, maintaining high adaptive performance even in long-term operation. Industrial applicability: It can be used in a wide range of applications, including corporate sales support, call center operations, outbound marketing, and automated response / calling systems using AI voice agents. It can also be applied to hybrid sales support, multi-channel customer support, virtual customer service, customer support bots, and interactive advertising response systems.

Claims

1. A cloud-based or on-premise CTI system comprising a customer information acquisition mechanism, an analysis engine, a voice mode database, and a voice conversion engine, The CTI system further includes a call control function that controls the outgoing and connecting of calls, The aforementioned customer information acquisition mechanism acquires customer characteristics from a storage device, including the customer's industry, company size, job title, past call results, and response history. The analysis engine calculates sales performance indicators corresponding to each of the multiple voice conversion modes based on the customer characteristics and call results or response history, and recommends the optimal voice conversion mode based on the sales performance indicators. The aforementioned sales performance indicators include at least one of the pass rate, response rate, and deal conversion rate. The aforementioned voice mode database stores voice feature parameters corresponding to multiple voice conversion modes, The voice conversion engine converts or generates operator voice or artificial intelligence-generated voice in real time based on the recommended voice conversion mode. The converted or generated audio is transmitted in accordance with the call processing performed by the call control function, characterized in that CTI system.

2. The CTI system further includes a learning mechanism, The learning mechanism feeds back the call results and response results to the analysis engine. The analysis engine is characterized by learning the correspondence between customer characteristics and sales performance indicators, and updating the recommendation accuracy of the voice conversion mode according to customer characteristics. The CTI system according to claim 1.

3. The learning mechanism learns the correspondence between the voice conversion mode and sales performance indicators for each customer characteristic based on the call results and response results. The analysis engine is updated to recommend a voice conversion mode that is expected to improve sales performance indicators for the customer characteristics in question. The CTI system according to claim 2.

4. The aforementioned CTI system is The recommended voice conversion mode can be applied to both operator-assisted calls and automated calls by artificial intelligence voice agents. The CTI system according to claim 1.

5. The aforementioned customer characteristics include the customer's industry classification. The aforementioned analysis engine is characterized by recommending different voice conversion modes for each industry classification. The CTI system according to claim 2.