阿里云声音复刻

阿里云声音复刻

个性化人声定制

阿里云个性化人声定制是智能语音交互产品自学习平台下的一部分

使用方式：https://help.aliyun.com/document_detail/456006.html

• 方式一：控制台界面定制使用方式

• 方式二：通过OpenAPI定制：在该页面有不同开发语言的示例代码，开发者使用openAPI概览注意只有四个API方法和声音克隆的API方法有所不同

Java示例代码

package com.alibaba.nls.ptts;

import com.aliyuncs.CommonRequest;
import com.aliyuncs.CommonResponse;
import com.aliyuncs.DefaultAcsClient;
import com.aliyuncs.IAcsClient;
import com.aliyuncs.exceptions.ClientException;
import com.aliyuncs.exceptions.ServerException;
import com.aliyuncs.http.MethodType;
import com.aliyuncs.http.ProtocolType;
import com.aliyuncs.profile.DefaultProfile;

public class PttsDemo {
    //域名
    private static final String DOMAIN = "nls-measure.cn-shanghai.aliyuncs.com";
    // API版本
    private static final String API_VERSION = "2019-09-05";

    private static IAcsClient client;

    static {
        String accessKeyId = "请替换为您自己的accessKeyId";
        String accessKeySecret = "请替换为您自己的accessKeySecret";
        // 创建DefaultAcsClient实例并初始化
        DefaultProfile profile = DefaultProfile.getProfile(
            "",
            accessKeyId,
            accessKeySecret);
        client = new DefaultAcsClient(profile);
    }

    public static void main(String[] args) throws InterruptedException {
        //20句音频，此处为了方便演示接口的调用，提前准备好了音频的url，在您集成时，可以在用户录制的过程中，将音频进行存储
        String[] urls = {"viwf/1.wav", "dycw/2.wav", "dopl/3.wav", "anfd/4.wav", "cyoy/5.wav", "dsjw/6.wav",
            "vevd/7.wav", "ulno/8.wav", "kwlw/9.wav", "lafu/10.wav", "uozh/11.wav", "gdpp/12.wav", "lisa/13.wav",
            "bmvv/14.wav", "ijzx/15.wav", "kdla/16.wav", "govf/17.wav", "kcid/18.wav", "srdx/19.wav", "stol/20.wav"};

        //训练的基本信息，voiceName请替换成您自己的命名
        String voiceName = "示例voice";
        String scenario = "story";
        String gender = "female";

        //Step1: 获取需要朗读的文本
        CommonRequest getDemonstrationRequest = buildRequest("GetDemonstrationForCustomizedVoice");
        getDemonstrationRequest.putQueryParameter("Scenario", scenario);
        String getDemonstrationResponse = sendRequest(getDemonstrationRequest);
        System.out.println("|获取需要朗读的内容|response=" + getDemonstrationResponse);

        //Step2: 采集用户朗读的音频，进行音频检测
        for (int i = 1; i <= 20; i++) {
            String audioUrl = audioRecordUrlPrefix + urls[i - 1];
            CommonRequest audioDetectRequest = buildRequest("CustomizedVoiceAudioDetect");
            audioDetectRequest.putQueryParameter("Scenario", scenario);
            audioDetectRequest.putQueryParameter("VoiceName", voiceName);
            audioDetectRequest.putQueryParameter("RecordUrl", audioUrl);
            audioDetectRequest.putQueryParameter("AudioRecordId", String.valueOf(i));
            String audioDetectResponse = sendRequest(audioDetectRequest);
            System.out.println("|音频检测|[" + i + "]response=" + audioDetectResponse);
        }

        //Step3: 20句音频检测完成后，提交训练
        CommonRequest submitTrainRequest = buildRequest("SubmitCustomizedVoice");
        submitTrainRequest.putQueryParameter("VoiceName", voiceName);
        submitTrainRequest.putQueryParameter("Gender", gender);
        submitTrainRequest.putQueryParameter("Scenario", scenario);

        String submitTrainResponse = sendRequest(submitTrainRequest);
        System.out.println("|提交训练|response=" + submitTrainResponse);

        //Step4: 轮询训练结果

        CommonRequest queryTrainResultRequest = buildRequest("ListCustomizedVoice");
        queryTrainResultRequest.putQueryParameter("VoiceName", voiceName);
        String queryTrainResultResponse = sendRequest(queryTrainResultRequest);
        System.out.println("|查询训练结果|response=" + queryTrainResultResponse);
    }

    private static String audioRecordUrlPrefix
        = "https://static-aliyun-doc.oss-cn-hangzhou.aliyuncs.com/file-manage-files/zh-CN/20221118/";

    private static CommonRequest buildRequest(String popApiName) {
        CommonRequest request = new CommonRequest();
        request.setSysMethod(MethodType.POST);
        request.setSysDomain(DOMAIN);
        request.setSysVersion(API_VERSION);
        request.setSysAction(popApiName);
        request.setSysProtocol(ProtocolType.HTTPS);
        return request;
    }

    private static String sendRequest(CommonRequest request) {
        try {
            CommonResponse response = client.getCommonResponse(request);
            return response.getData();
        } catch (ServerException e) {
            e.printStackTrace();
        } catch (ClientException e) {
            e.printStackTrace();
        }
        return null;
    }
}

阿里云个人化人声定制控制台

https://nls-portal.console.aliyun.com/pttssettingnew

非开发者使用指南：

https://help.aliyun.com/document_detail/451766.html?spm=a2c4g.432038.0.0.496f4906w8EQFD

声音克隆定制

阿里云产品帮助中心搜索声音克隆https://help.aliyun.com/?spm=a2c4g.456006.J_3207526240.2.6a5842e8QI7zZj

方法一：通过控制台进行人声克隆

1. 登录智能媒体服务控制台。
2. 在顶部左上角根据实际情况选择地域。
3. 在左侧导航栏选择****智能生产制作** > *人声克隆***。
4. 单击定制声音，根据提示完成定制人声操作。

方法二：通过OpenAPI进行人声克隆

1. 调用接口CreateCustomizedVoiceJob，配置参数VoiceId，创建人声克隆任务。

2. 调用接口GetDemonstrationForCustomizedVoiceJob，根据实际场景配置参数Scenario，获取需要朗读的20条示范文本及音频。

3. 根据步骤 2获取到的示范文本及音频进行训练并录制，并将录制的语音上传至OSS。上传方式，请参见上传文件。

4. 调用接口DetectAudioForCustomizedVoiceJob，提交已经训练好的音频，进行音频检测。

   重要 需要调用20次该接口，每次调用时提交对应的音频，共20条音频。

5. 调用接口SubmitCustomizedVoiceJob，提交人声克隆训练任务，系统开始进行语音训练。

API参考

人声克隆相关接口

openApI门户：https://next.api.aliyun.com/api/ICE/2020-11-09/CreateCustomizedVoiceJob?tab=DEMO&lang=JAVA

SDK开发包使用方法：https://help.aliyun.com/document_detail/262200.html?spm=a2c4g.378655.0.0.534f5068p0WEYq

运行实例注意区分SDK版本