The hottest one is under the scene of telephone re

2022-09-23
  • Detail

Immediately, finance independently developed ASR technology. In the recording scene, the customer's word accuracy rate reached 88%

ASR, that is, automatic speech recognition. This technology is through the conversion of voice to text, so that machines can understand people like ears. This technology has been widely used in many scenes, such as giving instructions through voice, and voice Notepad, voice intelligent toys and other products also include voice recognition and speech synthesis functions, as well as intelligent robots, intelligent outbound call systems, intelligent customer service systems and so on

in the actual application scenario in the field of call center, the application of automatic speech recognition technology is more challenging than the application of general PC system. For example, the complex noise environment, various dialects mixed, coupled with the rapid increase of vehicle ownership in different industries, there are a large number of proprietary sentence patterns and professional vocabulary, which makes speech recognition have technical problems and misunderstandings

it is understood that based on these pain points, instant consumer finance (hereinafter referred to as instant Finance) has created a self-developed speech recognition model, which can not only effectively solve the above problems, but also greatly improve the recognition accuracy. At present, the model has been implemented in many business scenarios of instant finance, such as intelligent outbound call, intelligent customer service, recording quality inspection, etc. According to the actual measurement, in the scene and various noise environments, the accuracy rate of instant financial ASR has reached 88%, which is at the leading level in the industry. Immediately, with its deep cultivation in the field of financial technology and the accumulation of vertical scene data, speech recognition technology has been applied in businesses such as smart double recording in the trust field and Chongqing department store

instant financial ASR technology has the ability to recognize multiple dialects. In order to meet the needs of different customer groups, the speech recognition technology of immediate finance continues to make efforts in the richness of languages. On the basis of the original common languages, the recognition ability of Sichuan dialect and Chongqing dialect has been opened, and the speech recognition ability of Cantonese and other dialects will be opened in succession. After long-term polishing training and effect optimization, Chongqing computer screen display experimental power and experimental curve speech recognition technology has been applied to business scenarios such as recording quality inspection in Chongqing department store

Dezhu intelligent speech quality inspection system is an intelligent speech quality inspection system that integrates speech recognition (ASR), voiceprint recognition, emotion recognition, self-developed language processing, but integrates a variety of advanced artificial intelligence (AI) technologies such as thin-walled plastic parts or very small gate (NLP). The system adopts advanced algorithms and makes full use of context information to improve the recognition effect of the model. At present, it has been applied to the scene of immediate financial call center, covering 58 nodes, 34 circulation paths and other highly complex businesses. The recall rate reaches 90% and the recognition accuracy rate reaches 86.5%

according to the 2019 data statistics of immediate finance, the detection rate of single violations of intelligent quality inspection of Dexi intelligent voice quality inspection system can be up to 36 times higher than that of manual sampling inspection. Among them, 7 intelligent quality inspection items are more than 5 times the efficiency of manual sampling inspection, the results of 14 intelligent quality inspection items are all better than manual sampling inspection, and the repayment efficiency of false promises of illegal quality inspection is increased by 13 times

Copyright © 2011 JIN SHI