twt-6ac3bc2a-bcad-460d-a344-43a386af587b.js

const AWS = require("aws-sdk");

var languageConfiguration = require("./language-configuration");

const polly = new AWS.Polly({
  signatureVersion: "v4",
  region: process.env.AWS_REGION,
});

const readTextFile = (filePath) => {
  const fs = require("fs");
  const text = fs.readFileSync(filePath, "utf-8");
  return text;
};

const synthesize = async (localFilePath) => {
  const text = readTextFile(localFilePath);
  const paragraphs = text.split("\n\n");
  const iso2Lang = "en";

  const audioBuffers = [];
  const langConfig =
    languageConfiguration.getLangConfigurationOrDefault(iso2Lang);

  for (const paragraph of paragraphs) {
    const paragraphWithBreak = `${paragraph} <break strength="x-strong" />`;

    const splittedText = chunkString(paragraphWithBreak, 1400);

    const paragraphBuffers = await Promise.all(
      splittedText.map((chunk) => {
        return polly
          .synthesizeSpeech({
            OutputFormat: "mp3",
            TextType: "ssml",
            Text: `<speak>${chunk}</speak>`,
            LanguageCode: langConfig.langCode,
            Engine: langConfig.engine,
            VoiceId: langConfig.voiceId,
          })
          .promise()
          .then((data) => data.AudioStream);
      })
    );

    audioBuffers.push(...paragraphBuffers);
  }

  const mergedBuffers = audioBuffers.reduce(
    (total, buffer) =>
      Buffer.concat([total, buffer], total.length + buffer.length),
    Buffer.alloc(1)
  );
  return mergedBuffers;
};

const chunkString = (paragraph, chunkSize) => {
  let splitString = [];
  for (let i = 0; i < paragraph.length; i = i + chunkSize) {
    splitString.push(paragraph.slice(i, i + chunkSize));
  }
  return splitString;
};

module.exports = {
  synthesize,
};