Spring Boot对接GCP语音转文本服务时遇身份验证错误

错误信息 #

Your default credentials were not found. To set up Application Default Credentials for your environment, see https://cloud.google.com/docs/authentication/external/set-up-adc.
    at com.google.auth.oauth2.DefaultCredentialsProvider.getDefaultCredentials(DefaultCredentialsProvider.java:127) ~[google-auth-library-oauth2-http-1.22.0.jar:1.22.0]
    at com.google.auth.oauth2.GoogleCredentials.getApplicationDefault(GoogleCredentials.java:152) ~[google-auth-library-oauth2-http-1.22.0.jar:1.22.0]
    at com.google.auth.oauth2.GoogleCredentials.getApplicationDefault(GoogleCredentials.java:124) ~[google-auth-library-oauth2-http-1.22.0.jar:1.22.0]
    at com.google.api.gax.core.GoogleCredentialsProvider.getCredentials(GoogleCredentialsProvider.java:70) ~[gax-2.42.0.jar:2.42.

核心代码 #

public String recognizeVoice(String encodedfile) throws IOException
{
    System.out.println("encoded file inside recognize Voice  "+ encodedfile);
    SpeechClient speechClient = SpeechClient.create(); // 报错行
    System.out.println("client created succesflly" );
    // The path to the audio file to transcribe
    String gcsUri = "src/main/resources/samplesSounds/punjabiAudioSpeech.wav";

    // Builds the sync recognize request
    RecognitionConfig config =
            RecognitionConfig.newBuilder()
                             .setEncoding(RecognitionConfig.AudioEncoding.LINEAR16)
                             .setSampleRateHertz(16000)
                             .setLanguageCode("pa-Guru-IN")
                             .setModel("command_and_search")
                             .setAudioChannelCount(2)
                             .build();

    RecognitionAudio audio = RecognitionAudio.newBuilder().setUri(gcsUri).build();

    // Performs speech recognition on the audio file
    RecognizeResponse response = speechClient.recognize(config, audio);
    List<SpeechRecognitionResult> results = response.getResultsList();

    for (SpeechRecognitionResult result : results) {
        // There can be several alternative transcripts for a given chunk of speech. Just use the
        // first (most likely) one here.
        SpeechRecognitionAlternative alternative = result.getAlternativesList().get(0);
        System.out.printf("Transcription: %s%n", alternative.getTranscript());
    }
    return "FAILED";
}

已尝试的配置 #

在Spring Boot的application.properties中配置：

spring.cloud.gcp.credentials.location=key.json

其中key.json为GCP服务账号密钥文件，但仍报错，询问其他访问该云服务的方式。

1. 显式加载凭据创建SpeechClient #

绕过默认凭据查找逻辑，直接在代码中加载服务账号密钥：

public String recognizeVoice(String encodedfile) throws IOException
{
    System.out.println("encoded file inside recognize Voice  "+ encodedfile);
    
    // 加载服务账号密钥（若密钥在resources目录下，使用getResourceAsStream）
    GoogleCredentials credentials = GoogleCredentials.fromStream(getClass().getResourceAsStream("/key.json"));
    SpeechSettings speechSettings = SpeechSettings.newBuilder()
            .setCredentialsProvider(FixedCredentialsProvider.create(credentials))
            .build();
    
    // 使用自定义配置创建SpeechClient
    SpeechClient speechClient = SpeechClient.create(speechSettings);
    System.out.println("client created succesflly" );
    
    // 后续语音识别代码保持不变...
}

2. 设置系统环境变量 #

通过环境变量指定凭据文件路径，让GCP SDK自动读取：

• Windows：命令行执行

  set GOOGLE_APPLICATION_CREDENTIALS=C:\绝对路径\key.json
  

• Linux/macOS：终端执行

  export GOOGLE_APPLICATION_CREDENTIALS=/绝对路径/key.json
  

设置完成后重启应用，SpeechClient.create()即可自动识别凭据。

3. 修正Spring Cloud GCP配置 #

若密钥文件放在src/main/resources目录下，需在配置中添加classpath:前缀：

spring.cloud.gcp.credentials.location=classpath:key.json

同时确保项目已引入Spring Cloud GCP语音服务依赖：

Maven依赖 #

<dependency>
    <groupId>com.google.cloud</groupId>
    <artifactId>spring-cloud-gcp-starter-speech</artifactId>
</dependency>

Gradle依赖 #

implementation 'com.google.cloud:spring-cloud-gcp-starter-speech'

4. 验证服务账号权限 #

确认GCP服务账号已拥有Cloud Speech-to-Text Admin或Cloud Speech-to-Text Editor权限，避免因权限不足导致认证失败。

内容的提问来源于stack exchange，提问作者Gurkirat Singh Guliani