Google Text-to-speech - 從 txt 文件的各行加載文本

Question

我在 Node.js 中使用 Google TextToSpeech API 從文本生成語音。 我能夠獲得與為語音生成的文本同名的 output 文件。 但是，我需要稍微調整一下。 我希望我可以同時生成多個文件。 關鍵是我有，例如，5 個單詞（或句子）要生成，例如 cat、dog、house、sky、sun。 我想將它們分別生成到一個單獨的文件中：cat.wav、dog.wav 等。

我還希望應用程序能夠從 *.txt 文件中讀取這些單詞（每個單詞/句子在 *.txt 文件的單獨行上）。

有這種可能嗎？ 下面我粘貼我正在使用的 *.js 文件代碼和 *.json 文件代碼。

*.js

const textToSpeech = require('@google-cloud/text-to-speech');
const fs = require('fs');
const util = require('util');
const projectId = 'forward-dream-295509'
const keyFilename = 'myauth.json'
const client = new textToSpeech.TextToSpeechClient({ projectId, keyFilename });
const YourSetting = fs.readFileSync('setting.json');
async function Text2Speech(YourSetting) {
  const [response] = await client.synthesizeSpeech(JSON.parse(YourSetting));
  const writeFile = util.promisify(fs.writeFile);
  await writeFile(JSON.parse(YourSetting).input.text + '.wav', response.audioContent, 'binary');
  console.log(`Audio content written to file: ${JSON.parse(YourSetting).input.text}`);
}
Text2Speech(YourSetting);

*.json

{
  "audioConfig": {
    "audioEncoding": "LINEAR16",
    "pitch": -2,
    "speakingRate": 1
  },
  "input": {
    "text": "Text to Speech" 
  },
  "voice": {
    "languageCode": "en-US",
    "name": "en-US-Wavenet-D"
  }
}

我不太擅長編程。 我在 google 上找到了一個關於如何執行此操作的教程，並對其進行了一些修改，以便保存文件的名稱與生成的文本相同。

我將非常感謝您的幫助。 阿雷克

Answer 1

這里是 go - 我還沒有測試過，但這應該顯示如何讀取文本文件，分成每一行，然后以一組並發運行 tts。 它使用您需要添加到項目中的 p-any 和 filenamify npm 包。 請注意，谷歌可能有 API 限制或我在這里沒有考慮的速率限制 - 如果這是一個問題，可以考慮使用 p-throttle 庫。

// https://www.npmjs.com/package/p-map
const pMap = require('p-map');

// https://github.com/sindresorhus/filenamify
const filenamify = require('filenamify');


const textToSpeech = require('@google-cloud/text-to-speech');
const fs = require('fs');
const path = require('path');
const projectId = 'forward-dream-295509'
const keyFilename = 'myauth.json'
const client = new textToSpeech.TextToSpeechClient({ projectId, keyFilename });
const rawSettings = fs.readFileSync('setting.json', { encoding: 'utf8'});

// base data for all requests (voice, etc)
const yourSetting = JSON.parse(rawSettings);

// where wav files will be put
const outputDirectory = '.';

async function Text2Speech(text, outputPath) {
    // include the settings in settings.json, but change text input
    const request = {
        ...yourSetting,
        input: { text }
    };
    const [response] = await client.synthesizeSpeech(request);

    await fs.promises.writeFile(outputPath, response.audioContent, 'binary');
    console.log(`Audio content written to file: ${text} = ${outputPath}`);
    // not really necessary, but you could return something if you wanted to
    return response;
}

// process a line of text - write to file and report result (success/error)
async function processLine(text, index) {
    // create output path based on text input (use library to ensure it's filename safe)
    const outputPath = path.join(outputDirectory, filenamify(text) + '.wav');
    const result = {
        text,
        lineNumber: index,
        path: outputPath,
        isSuccess: null,
        error: null
    };
    try {
        const response = await Text2Speech(text, outputPath);
        result.isSuccess = true;
    } catch (error) {
        console.warn(`Failed: ${text}`, error);
        result.isSuccess = false;
        result.error = error;
    }
    return result;
}

async function processInputFile(filepath, concurrency = 3) {
    const rawText = fs.readFileSync(filepath, { encoding: 'utf8'});
    const lines = rawText
        // split into one item per line
        .split(/[\r\n]+/)
        // remove surrounding whitespace
        .map(s => s.trim())
        // remove empty lines
        .filter(Boolean);
    
    const results = await pMap(lines, processLine, { concurrency });
    console.log('Done!');
    console.table(results);
}

// create sample text file
const sampleText = `Hello World
cat
dog
another line of text`;
fs.writeFileSync('./my-text-lines.txt', sampleText);

// process each line in the text file, 3 at a time
processInputFile('./my-text-lines.txt', 3);