Voice recognition combines voice print detection and text review capabilities to effectively identify pornographic, wheezing, illegal, abusive and other illegal voice content, and supports short audio real-time detection, long audio and audio stream asynchronous detection and other methods.Help customers such as audio and video live broadcast, radio station and chat room to create a green and clear voice environment
Based on deep learning algorithm and massive data accumulation, build a special model for live broadcast scenes to support accurate recognition of voice content in the presence of background noise
Yellow related content identification
Accurately identify pornographic behavior descriptions, vulgar friends and other pornographic content in audio
Squeezing sound recognition
Detect and identify wheezing, panting, groaning and ASMR contained in audio
Identification of prohibited and illegal content
Identify violence, terrorist descriptions, gambling, drugs, guns and ammunition and other prohibited content in voice content
Identification of vulgar abusive content
Recognize abusive, personal attacks, negative catharsis and other harmful content in voice content
Identification of malicious promotion content
Identify illegal advertisements such as illegal products in voice content, personal contact information such as WeChat/QQ and other illegal content
Advertising law review
Identify the illegal words prohibited by advertising law in voice content and update them regularly
Self defined identification of black and white list
Black and white lists can be customized for business scenarios to meet business needs
Application scenarios
Live audio
Audio on demand
Voice chat
Online education
Live audio
Audio and video live broadcast scenes, review voice content such as anchor speech and singing, accurately detect illegal content such as pornography and abuse, and help the platform create a green live broadcast environment
Access mode
Short audio synchronization audit
Quickly detect audio files within 60 seconds, and synchronously return audit results in real time, which is suitable for scenarios with high timeliness requirements
Learn more
Audio file asynchronous audit
Batch detect audio files within 60 minutes, and asynchronously return audit results, which is suitable for batch detection scenarios of a large number of tasks
Learn more
Audio stream audit
Real time parsing of audio stream links, asynchronous segmentation detection and return of audit results, applicable to live broadcast scenarios
Learn more
product price
Free version
Short Audio - Pay As You Go
Short Audio - Prepaid
Audio files - pay as you go
Audio files - prepaid
Custom Edition
Free version
After completing the real name authentication, you can enjoy free testing resources, which are suitable for enterprises/individuals in the testing phase.Free use>
Technical features
High recognition accuracy
Special optimization model for audio and video live broadcast scenes, accurate recognition of voice content, multi-dimensional and efficient text review, extensive application of multi-disciplinary violation review, significantly improving the efficiency of enterprise review
Super sensitive thesaurus
Ten million level sensitive thesaurus, with data updated every day, continuously improving the effect of audit capability, can quickly identify new complex illegal texts
Support free customization
According to the personalized data requirements of business scenarios, the black-and-white list text thesaurus can be freely customized, and the threshold parameters of the audit model can be adjusted as needed to flexibly optimize the recognition effect
Experience AI intelligent voice audit capability for free
Public cloud API can enjoy up to 2000 free test resources