bililive-touhou-tts/tts_bridge.js

/** tts_bridge.js — Persistent Node.js bridge for aquestalk.js TTS synthesis.

Converts English/romaji to katakana via wanakana before synthesis.
*/

import { readFileSync, writeFileSync } from "fs";
import { createInterface } from "readline";
import { fileURLToPath, pathToFileURL } from "url";
import { dirname, join } from "path";
import wanakana from "wanakana";

const __filename = fileURLToPath(import.meta.url);
const __dirname = dirname(__filename);

const args = process.argv.slice(2);
let voice = "f1";
let speed = 100;

for (let i = 0; i < args.length; i++) {
  if (args[i] === "--voice" && i + 1 < args.length) {
    voice = args[++i];
  } else if (args[i] === "--speed" && i + 1 < args.length) {
    speed = parseInt(args[++i], 10);
  }
}

const aquestalkMod = pathToFileURL(join(__dirname, "aquestalk.js", "dist", "index.js")).href;
const { load } = await import(aquestalkMod);

console.error(`[bridge] Loading aquestalk.js with voice="${voice}", speed=${speed}...`);

let aq;
try {
  aq = await load(voice);
  console.error("[bridge] aquestalk.js loaded successfully.");
} catch (err) {
  console.error("[bridge] Failed to load aquestalk.js:", err.message);
  process.stdout.write("ERR:FAILED_TO_LOAD\n");
  process.exit(1);
}

function synthesize(kanaText) {
  const trimmed = kanaText.trim();
  if (!trimmed) throw new Error("EMPTY_TEXT");
  const finalText = wanakana.toKatakana(trimmed, { convertLongVowelMark: true });
  const wav = aq.run(finalText, speed);
  return Buffer.from(wav);
}

process.stdout.write("READY\n");

const rl = createInterface({
  input: process.stdin,
  output: process.stdout,
  terminal: false,
});

rl.on("line", (line) => {
  line = line.trim();
  if (!line) return;
  const sepIdx = line.indexOf("|");
  if (sepIdx === -1) {
    process.stdout.write(`ERR:INVALID_FORMAT:${line}\n`);
    return;
  }
  const inputPath = line.substring(0, sepIdx);
  const outputPath = line.substring(sepIdx + 1);
  try {
    const kanaText = readFileSync(inputPath, "utf-8").trim();
    if (!kanaText) {
      process.stdout.write(`ERR:EMPTY_TEXT\n`);
      return;
    }
    const wav = synthesize(kanaText);
    writeFileSync(outputPath, wav);
    process.stdout.write(`OK:${outputPath}\n`);
  } catch (err) {
    process.stdout.write(`ERR:${err.message}\n`);
  }
});

rl.on("close", () => {
  console.error("[bridge] stdin closed, shutting down.");
  aq.destroy().then(() => process.exit(0));
});