tencent cloud

腾讯云智能数智人

产品简介
产品概述
产品功能
产品优势
购买指南
价格指南
购买指引
使用代金券购买流程
退费说明
数智人形象介绍
形象类别介绍
基础形象库
形象和声音复刻指引
数智人平台操作指引
访问平台
形象生产与资产管理
数智人会话互动应用与管理
播报数智人视频生成与管理
运营管理与分析
服务端 API 接入
数智人 API 接入模式概述
数智人 aPaas 接口调用方式
2D真人小样本形象、声音定制 API 文档
视频生成服务 API 文档
交互数智人服务 API 文档
个人资产管理 API 文档
客户端 SDK 接入
整体介绍
3D 端渲染 SDK 接入
2D 端渲染 SDK 接入
数智人 SSML 标记语言规范
相关协议
隐私协议
DSA(Data Sharing Agreement)
常见问题
文档腾讯云智能数智人

Custom Image or Voice Practical Tutorial

聚焦模式
字号
最后更新时间: 2025-11-05 10:20:16

Scenarios

This document provides end-to-end operation guidelines for 2D real-person avatar and Customized Text To Speech (CTTS).

Prerequisites

Official Environment Primary Domain Name: https://gw-sg.tvs.qq.com

Custom Avatar

Step One: Obtain a Temporary Token and Upload the Material File to Tencent Cloud Storage Service (COS)

Target:
Upload the material files to the cloud storage service using the temporary token for future reference by the avatar customization service to obtain the material files. The process mainly involves two steps.
Step 1:
Obtain the temporary token and related parameters. Please refer to Obtain the Temporary Upload Token to get the following four key pieces of information: Token, TmpSecretId, TmpSecretKey, and PathPrefix. Then, parse out more parameter information from the obtained PathPrefix in the format PathPrefix=https://{bucket}.cos.{region}.myqcloud.com/{key}. Finally, this step can obtain the following parameters: Token (one of the three elements of the temporary key), TmpSecretId (one of the three elements of the temporary key, SecretId), TmpSecretKey (one of the three elements of the temporary key, SecretKey), Bucket (bucket name), Region (storage service region), and Key (component of the bucket object's domain names or IP addresses).
Step 2:
Call the COS_SDK to upload material files to Tencent Cloud Storage Service (COS). Please refer to Uploading Materials to Tencent Cloud COS. Choose the appropriate SDK according to the actual situation. The required parameters in this SDK include: TmpSecretId, TmpSecretKey, TmpToken, region, coskey, and local file address.
TmpSecretId: Use the TmpSecretId obtained in the previous step.
TmpSecretKey: Use the TmpSecretKey obtained in the previous step.
TmpToken: Get the token obtained in the previous step.
Region: Use the Region obtained in the previous step.
CosKey: Concatenate the key obtained in the previous step. There are 5 following cases.
Upload video authorization, coskey={key}idcard/xxxx.mp4
Upload written authorization letter, coskey={key}idcard/xxxx.pdf
Upload video footage for custom avatar, coskey={key}video/xxxx.mp4
Upload photo materials for custom avatar, coskey={key}photo/xxxx.png (support jpg/jpeg/png/webp)
Upload audio materials for custom voice, coskey={key}audio/xxxx.zip (for package content and format requirements, see the MaterialCosUrl parameter introduction in custom API).
Local address of material file: fill in according to the actual situation just.
Result returned:
After requesting via SDK, you will receive a prompt indicating the material upload succeeded and the corresponding material link. You can click View to check whether the upload was successful.

Step Two: Call the Custom API to Customize Image or Voice

Target:
Through this API, initiate customized service. The program will automatically pull materials from the previous step for customization.
Note:
Purchase the corresponding avatar or voice type customization quota in advance.
For specific calls, please refer to Customization API. The parameter values of IdentityCosUrl, IdentityWrittenCosUrl, and MaterialCosUrl are equal to "/{coskey}" (coskey refers to the parameter in step one).
Returned result:
The custom task will return a unique TaskId. Subsequent steps will query the status and download based on this TaskId.

Step 3: Query the Customization Progress of Image or Voice Via the Query API

Target:
Use this API to get the progress and results of the custom task in real time.
For specific calls, please refer to Progress Query API.
The result returned:
Query StageInfo data content based on different status. When customization is complete, you can query the Demo address for effect confirmation. Likewise, you can query sound.

Step Four: Effect Confirmation

Target:
After confirming via the effect confirmation API, the avatar will be assigned to the designated UIN account and used for session interaction or video generation.
For specific calls, please refer to Effect Confirmation API.

Voice replication

On the basis of a custom avatar, a front audio quality inspection work will be added, mainly divided into two scenarios: basic version and ultra-fast version.

Basic Version

Prerequisites: Detect the submitted sound quality. See audio quality testing task creation API.
Preparation 2: Perform status query for the created audio quality inspection task. See audio quality inspection task status query interface.
Follow up by referring to the custom avatar process to advance.

Ultra-Fast Version

Preparation 1: Call the API to obtain the designated reading text. Please see Obtaining the Voice Training Text.
preparations 2: record audio manually.
Preparation 3: Detect the submitted sound quality. See submit audio quality inspection task.
Preparation 4: Perform status query for the created audio quality inspection task. See query audio quality inspection task progress.
Refer to the custom avatar process and advance subsequently.



帮助和支持

本页内容是否解决了您的问题?

填写满意度调查问卷,共创更好文档体验。

文档反馈