Speech Recognition Engine

Support dynamic language model, voice service customization

Speech Recognition Engine

Automatic Speech Recognition , a service used to recognize the user's voice and translate the voice into text. Its goal is to convert the vocabulary content in human speech into computer-readable input, so that the device has the ability to 'hear' and realize human-computer interaction.

Speech Recognition Engine

Features

行业痛点

功能

Advantage

行业痛点

支持多语言多口音
支持多语言多口音

先进的算法技术,支持CTC+Transformer


精准识别率
精准识别率

资源占用灵活配置,实时率高达0.1-0.15


​超大规模模型训练
​超大规模模型训练

采用神经网络对标点符、数字转换等可读性模块预测,多种机器学习算法的融合

支持动态语言模型
支持动态语言模型

定制化训练,私有化部署,能够快速对深度神经网络进行计算,从而加快解码速度

Related

Voice Analysis System

Realize second-level analysis, with the highest recognition rate in insurance, banking and other fields

Real-time Agent Assistant

Speaking tips and warnings, restraining non-compliant behaviors of agents

Voice Quality Inspection System

Daily processing capacity 1000 hours+, 100% full quality inspection

Intelligent Voice Robot

Support two scenarios of outbound and inbound calls, human-machine collaboration

Visual Analysis System

Visualized real-time large screen based on voice big data, real-time public opinion warning

Car Voice Assistant

Provide end users with an integrated “cloud + terminal + core” automotive intelligent interactive service

Smart Training

Support multi-channel teaching such as APP and mini programs, and teach speech in a scene interactive way

AI Capability Platform

Deep integration of AI based on the two core capabilities of speech recognition and speech synthesis