RESOURCE_PROJECT_INVALID 错误：GCP 语音转文本 v2 NodeJS

Question

我正在尝试在 NodeJS 应用程序中使用 Cloud Speech-To-Text v2。我的代码与 v1 配合良好：

import { v1 } from '@google-cloud/speech';
const { SpeechClient } = v1;
let speechClient: v1.SpeechClient | undefined;

speechClient = new SpeechClient(); // GOOGLE_APPLICATION_CREDENTIALS is set to ./google-credentials.json, which exists, is valid and has the Roles 'Cloud Speech CLient', Cloud Speech-to-Text Service Agent' and 'Cloud Translation API User'

const googleCloudSpeechConfig = {
    config: {
        encoding: "LINEAR16",
        sampleRateHertz: 16000,
        languageCode: "EN-US"
    },
    interimResults: true,
};
recognizeStream = speechClient
    .streamingRecognize(config)
    .on('error', (err) => { ... })
    .on('data', async (data) => { ... });

但是，当我使用

切换到 v2 时

import { v2 } from '@google-cloud/speech';
const { SpeechClient } = v2;
let speechClient: v2.SpeechClient | undefined;

我收到错误

Error: 3 INVALID_ARGUMENT: Invalid resource field value in the request.
at callErrorFromStatus (/Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/@grpc/grpc-js/src/call.ts:82:17)
at Object.onReceiveStatus (/Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/@grpc/grpc-js/src/client.ts:705:51)
at Object.onReceiveStatus (/Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/@grpc/grpc-js/src/client-interceptors.ts:419:48)
at /Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/@grpc/grpc-js/src/resolving-call.ts:132:24
at processTicksAndRejections (node:internal/process/task_queues:77:11)
for call at
at ServiceClientImpl.makeBidiStreamRequest (/Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/@grpc/grpc-js/src/client.ts:689:42)
at ServiceClientImpl.<anonymous> (/Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/@grpc/grpc-js/src/make-client.ts:189:15)
at /Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/@google-cloud/speech/build/src/v2/speech_client.js:317:29
at /Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/google-gax/build/src/streamingCalls/streamingApiCaller.js:46:28
at /Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/google-gax/build/src/normalCalls/timeout.js:44:16
at StreamProxy.setStream (/Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/google-gax/build/src/streamingCalls/streaming.js:144:24)
at StreamingApiCaller.call (/Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/google-gax/build/src/streamingCalls/streamingApiCaller.js:54:16)
at /Users/dominik/Documents/Git Projects/event_subtitles/server/node_modules/google-gax/build/src/createApiCall.js:84:30
at processTicksAndRejections (node:internal/process/task_queues:95:5) {
code: 3,
details: 'Invalid resource field value in the request.',
metadata: Metadata {
internalRepr: Map(2) {
'google.rpc.errorinfo-bin' => [Array],
'grpc-status-details-bin' => [Array]
},
options: {}
},
statusDetails: [
ErrorInfo {
metadata: [Object],
reason: 'RESOURCE_PROJECT_INVALID',
domain: 'googleapis.com'
}
],
reason: 'RESOURCE_PROJECT_INVALID',
domain: 'googleapis.com',
errorInfoMetadata: {
service: 'speech.googleapis.com',
method: 'google.cloud.speech.v2.Speech.StreamingRecognize'
}
}

怎么会？

我还注意到这些类型看起来很奇怪。

v2.SpeechClient.streamingRecognize

期望类型

streamingConfig?: google.cloud.speech.v1.IStreamingRecognitionConfig | google.cloud.speech.v1p1beta1.IStreamingRecognitionConfig | undefined

这看起来很奇怪。

此外，文档位于

https://cloud.google.com/speech-to-text/v2/docs/streaming-recognize

有 Python 示例，但没有 NodeJS 示例。

我做错了什么？

我知道这是重复的 如何设置流式传输在 Node.js 中识别 Google Cloud Speech To Text V2？

但是那里没有有效的答案。有这样的： https://github.com/GoogleCloudPlatform/nodejs-docs-samples/issues/3578

但它对我不起作用，我收到错误：3 INVALID_ARGUMENT，正如另一个用户也报告的那样。

Answer 1

我终于让它发挥作用了。我遵循 TypeScript 定义提供的类型来理解所需的嵌套配置结构。如果您使用 JS 代码，只需省略类型即可。

首先要注意三件事：

1. 在 v2 中，如果不使用“识别器”（这是预先保存的配置），他的功能似乎无法工作。不过我们可以使用默认识别器。
2. 您没有将配置传递给 _streamingRecognize()。相反，您将其作为第一次写入发送，然后将音频作为对象 { audio: data }
3. 标准的streamingRecognize()不起作用。它会产生上述错误。相反，请使用 _streamingRecognize()。

这是我的有效代码：

const recognitionConfig: google.cloud.speech.v2.IRecognitionConfig = {
        autoDecodingConfig: {},
        explicitDecodingConfig: {
            encoding: event.encoding,
            sampleRateHertz: event.sampleRateHertz,
            audioChannelCount: 1,
        },
        languageCodes: [event.languageCode],
        model: 'long' // video does not exist in v2
    }

    const streamingRecognitionConfig: google.cloud.speech.v2.IStreamingRecognitionConfig = {
        config: recognitionConfig,
        streamingFeatures: {
            interimResults: true,
        }
    }

    const streamingRecognizeRequest: google.cloud.speech.v2.IStreamingRecognizeRequest = {
        recognizer: `projects/${GOOGLE_PROJECT_ID}/locations/global/recognizers/_`,
        streamingConfig: streamingRecognitionConfig,
    };

recognizeStream = speechClient
    ._streamingRecognize()
    .on('error', (err) => {
        console.error(err);
    })
    .on('data', async (data) => { ... });
recognizeStream.write(streamingRecognizeRequest); // Do this once and only once

发送音频垃圾时，一定要发送

recognizeStream.write({ audio: data }); // where data is your audio chunk

记下 GOOGLE_PROJECT_ID，您在其中放置项目 ID。您可以在 Google Cloud Console 中找到它。

现在关于识别器 URL - 如果我使用其他区域，调用会失败。我怀疑你必须首先创建一个识别器才能做到这一点。更多内容请参阅本期。