Tencent Cloud Automatic Speech Recognition (ASR) provides highly cost-effective speech recognition services. It has been widely used by many Tencent businesses such as WeChat, Honor of Kings, and Tencent Video and has implemented multiple use cases, including recording quality inspection, real-time meeting transcription, and voice input method.
Features
Real-time speech recognition
It recognizes real-time audio streams to achieve the effect of instant speech-to-text, which is suitable for real-time audio streaming scenarios such as voice input and phone bot.
Recording file recognition
Recognizes recording files and allows asynchronous processing of lengthy audio recordings, applicable to long audio scenarios such as customer service quality inspection and subtitle generation.
Strengths
Massive data accumulation
Based on Tencent's vast social data platform, ASR has accumulated hundreds of thousands of hours of annotated voice data in a rich and diverse corpus, laying a data foundation for a high recognition accuracy.
Industry-leading algorithms
Based on multiple sequential neural network structures (LSTM, Attention Model, and DeepCNN), ASR is trained in the multitask learning method and delivers an industry-leading recognition accuracy together with the T/S approach in general and vertical fields.
Cross-platform Support
ASR provides RESTful APIs and SDKs and supports a wide variety of devices and terminals, including smart hardware, mobile application, website, desktop client, and IoT.
Support for Multiple Languages
ASR currently supports speech recognition in Mandarin and English, with more languages to come in the future.
Excellent recognition performance in noisy environment
ASR features robust recognition models, high recognition accuracy, and strong noise resistance. It can recognize audio information from noisy environments with no need of noise reduction processing.
Well proven capabilities
ASR has been fully verified by Tencent's internal businesses such as WeChat, Tencent Video, and Honor of Kings and has implemented many external use cases for customers in the internet, finance, education, and other industries, serving billions of users every day with a stable performance.
Application Scenario
Voice input method
ASR makes smart voice input possible through real-time speech recognition, which saves users the input time and improves the input experience.
Meeting Minutes
Audio information in conferences, court trials, and interviews can be converted to text by the real-time speech recognition service, which reduces human recording costs and improves the efficiency.
Call quality inspection
Rep conversations can be converted to text by the real-time speech recognition service, which comprehensively covers the content and improves the efficiency of quality inspection.