tencent cloud

Cloud Streaming Services

Release Notes and Announcements
Announcements
User Guide
Product Introduction
Overview
CSS Products
Concepts
Features
Use Cases
Strengths
Use Limits
Purchase Guide
Billing Overview
Basic Service Fee
Value-Added Service Fee
Prepaid plan
Purchase Process
Changing Billing Modes
Refund Policy
Viewing Bills
Renewal
Processing for Overdue Payments
Billing FAQs
Live Video Broadcasting (LVB)
Overview
Use Cases
Getting Started
SDK Integration
Live Event Broadcasting (LEB)
Overview
LEB Versus LVB
Use Cases
Getting Started
SDK Integration
Live Video Caster
Overview
Application Scenarios
Feature Area Introduction
Managing Live Video Caster
General Cloud Director
Configuring Program Lists and Automated Broadcasting
Console Guide
Console Overview
Overview
Domain Management
Stream Management
Package Management
AI Features
Feature Configuration
Relay
Billing Usage Statistics
Monitoring
Toolkit
OOTB live
CAM-Based Access Control
Feature Guide
Push and Playback
Features
Practices in Typical Scenarios
Cloud Native Recording
Live Streaming Security
Global CSS Service
Callback Notifications
User Guides for Common Third-Party Tools
SDK Guide
0. SDK Integration Guide
1. Stream Push
2. Playback
3. Advanced Features
API Documentation
History
Introduction
API Category
Making API Requests
Live Pad APIs
Live Stream Mix APIs
Time Shifting APIs
Monitoring Data Query APIs
Billing Data Query APIs
Live Transcoding APIs
Delayed Playback Management APIs
Domain Name Management APIs
Watermark Management APIs
Certificate Management APIs
Stream Pulling APIs
Recording Management APIs
Live Callback APIs
Screencapturing and Porn Detection APIs
Authentication Management APIs
Live Stream Management APIs
Data Types
Error Codes
Ops Guide
Video Stuttering
Troubleshooting Push Failure
Troubleshooting Playback Failure
CLS Assists in Live Stream Troubleshooting
Troubleshooting High Latency
Troubleshooting Poor Quality of Pulled Video
Authorizing CSS to Store Screenshots in a COS Bucket
Troubleshooting
Live Stream Mixing Error `InvalidParameter.OtherError`
About Pushing
Generating Push URLs
PC Push
Playing Method
Web Player
Live Streaming Quiz
FAQs
Service Region
Live Streaming Basics
Push and Playback
CSS Billing
Global CSS Service
Live Recording
On-Cloud Stream Mix
Domain Configuration
Related to Live Video Caster
Compatibility with Apple ATS
Difference Between Stream Interruption and Stream Suspension
SLA
CSS Service Level Agreement
CSS Policy
Privacy Policy
Data Processing And Security Agreement
Glossary

Moderation Templates

PDF
Focus Mode
Font Size
Last updated: 2025-12-03 18:07:10
CSS offers a live stream moderation feature. You can configure moderation templates in the console. Once the streaming domain is associated with the template, this service will obtain live screenshots and the audio during streaming, storing any non-compliant screenshots or audio data in Tencent Cloud COS. If the streaming domain is linked to the callback configuration, Tencent Cloud will send a request to the customer's server once a callback event is triggered during the live stream. Customer's server is responsible for handling the request. After verification, a JSON packet containing the moderation callback information can be obtained. This document provides instructions on how to create, bind, unbind, modify, and delete moderation configuration templates via the console.

Notes

Live stream moderation is a paid feature. Image moderation is charged 0.2294 USD per thousand images, while audio moderation is charged 0.0021 USD per minute. For more information, please refer to Live Stream Moderation.For more information, see Value-Added Services - Live Stream Moderation.
Screenshots or audio flagged during live stream moderation are stored in Tencent Cloud COS, which will generate storage costs. For pricing details, please refer to COS Product Pricing.
If your COS bucket allows public read access and has politically sensitive, pornographic, or other inappropriate content, to avoid the bucket being blocked, please delete the content first.
After a template is created, it can be bound with a push domain. For more information, see Live Stream Moderation Configuration. The association of the template is usually effective within 5-10 minutes.
Binding, modifying, or unbinding a template only affects new live streams and not ongoing ones. To apply new rules to ongoing live streams, you need to stop them and push them again.
To use a custom keyword library for image recognition, make sure "OCR" of "Image recognition" is selected. When OCR is on, the system will moderate the text in video screenshots. OCR-based moderation detects all non-compliant content regardless of the recognition policy configured.
Smart moderation can be performed on the live stream image, audio, and text content. Users can customize moderation policies and flexibly configure identification dimensions to ensure live streaming content compliance.

Prerequisites

You have activated the Tencent CSS service and added a push domain.
You have created a COS bucket. For more details, see Creating a Bucket.

Creating Live Stream Moderation Templates

1. Log in to the CSS console and choose Feature Configuration > Content moderation from the left navigation bar.
Note:
CSS stores video screenshots and audio data in COS buckets. Therefore, if it's your first time creating a moderation template, you need to create a service role and grant CSS read and write access to COS.
2. Click Create template.
3. A pop-up window will appear for you to grant CSS authorization to access COS resources. Click Authorize Now to go to the Role Management page.



4. On the Role Management page, click Grant. After authentication, CSS will have access to COS resources, and you can start using the live stream moderation feature.



5. After authorization, you will be redirected to the Create template page, where you can fill in the configuration items.

Steps

Step 1: Configuring Moderation

Click Create template to configure moderation:

Configuration Item
Description
Template Name
The template name can include up to 30 Chinese characters, letters, digits, underscores (_), and hyphens (-).
Template Description
The template description can include up to 100 Chinese characters, letters, digits, spaces, underscores (_), and hyphens (-).

Step 2: Configuring Screencapture/Segment Policy

In Screencapture/Segment policy, you can configure what content you want to moderate.
Configuration Item
Required Item
Description
Content type to moderate
Yes
Screenshot images
Screencapture interval: The interval at which screenshots are taken during a live stream. The shorter the interval, the finer the moderation granularity, and the higher the cost.
The screencapture interval can be 2-300 seconds. The default value is 2 seconds.
Audio
The audio moderation feature only supports Chinese audio.
Audio segment duration refers to the length of each audio segment extracted from a live stream. It determines the duration of the non-compliant audio segments returned and does not affect billing.
The audio segment duration can be 15-60 seconds. The default value is 15 seconds.
Audio text recognition
Audio text recognition is only used for features such as smart erase but will not trigger review.The Smart Erase feature utilizes its configured policies to identify and eliminate audio content that violates regulations.


Step 3: Configuring Recognition Policy

1. On the recognition policy configuration page, you can configure policies for both Image Recognition and Audio Recognition.
2. You can expand each category of Image Recognition and Audio Recognition to select sub-categories for moderation.
Note:
The system will recognize and perform moderation on the content according to the categories you select. If no categories are selected, no recognition or moderation will be performed.
To use a custom keyword library for image recognition, make sure "OCR" of "Image recognition" is selected.
Image Recognition Configuration
Audio Recognition Configuration
Audio text recognition configuration



3. From the drop-down list of custom keyword libraries, you can select a custom library to use for image recognition or audio recognition.



Note:
If you need to use a custom keyword library for content recognition, you need to configure one in the console first. For more information, see Custom Keyword Libraries.
4. Configure the storage path for saving screenshots or audio for moderation. Then, click Save to save the moderation template.

Configuration Item
Required Item
Description
Storage Location
_
The screenshot will be stored in your configured COS bucket, and make sure that the COS bucket has authorized CSS write request. Please create a COS bucket and authorize, reference documentation.
Storage path
Yes
Choose a COS bucket that you have created and authorized in COS.
Region refers to the geographic information of the selected bucket, which cannot be edited.
Backup storage path
No
Backup storage is supported. If the primary bucket does not work properly, the moderation content will be automatically stored in the backup bucket. The backup bucket will be used only if the primary bucket is down, and only one copy of data will be stored. The primary Region and backup Region cannot be identical.
Folder
Yes
Click the input box to select a COS folder. The default is/Audit/{Year}-{Month}-{Day}/. Note: COS folder names only allow [a-z, A-Z, 0-9] characters and the symbols -,!, _,., * along with placeholders.
File Name
Yes
The file name format can be customized through parameter assembly. Default is:
{StreamID}-Audit-{Hour}-{Minute}-{Second}{Ext}, wherein:
{StreamID}: Stream ID
{Audit}: Moderation
{Hour}: Moderation time (Hour)
{Minute}: Moderation time (Minute)
{Second}: Moderation time (Second)
{Ext}: Extension (.jpg)
Note: It only accepts [a-z, A-Z, 0-9], symbols -,!,_,.,*, and placeholders.
For example, if you enter the file format as {Hour}-{Minute}-{Second}-{Ext}, a screenshot taken during a live stream at 14:00:00 will be stored in COS with the filename 140000.jpg.

Binding Domain Names

1. Log in to the CSS console and choose Feature Configuration > Content moderation from the left navigation bar.
2. You can bind the moderation template to a push domain with the following methods:
Directly bind a domain: Click Bind Domain Name in the top left.

Bind a domain after creating a new moderation template: After successfully creating a moderation template, click Bind Domain Name in the pop-up window.

3. In the Bind Domain Name window, select the moderation template and push domain name(Multiple push domain names can be bound simultaneously) you want to bind together, and click Confirm to bind them.


Unbinding

1. Log in to the CSS console and choose Feature Configuration > Content moderation from the left navigation bar.
2. Select the live stream moderation template from which you want to unbind the push domain and click Unbind.

3. In the pop-up window, click Confirm.




Modifying a Template

1. From the left navigation bar, choose Feature Configuration > Content moderation.
2. Select the successfully created moderation template, click Edit on the right to go to modify the template information.


Deleting a Template

Note:
If the template is already bound to a push domain, you must first unbind it before the template can be deleted.
A deleted template cannot be recovered. Proceed with caution when deleting templates.
1. From the left navigation bar, choose Feature Configuration > Content moderation.
2. Select a previously created moderation template and click Delete.

3. Click Confirm to permanently delete the template.


Related Operations

For detailed instructions and additional information on binding and unbinding moderation templates at the domain level, please refer to Moderation Configuration.


Help and Support

Was this page helpful?

Help us improve! Rate your documentation experience in 5 mins.

Feedback