feat: Bilibili Live Danmaku -> Yukkuri TTS Reader
- Python main entry with CLI (--room-id, --voice, --speed) - blivedm integration for real-time danmaku receiving - Chinese-to-Japanese katakana conversion (pypinyin + mapping table) - Node.js persistent bridge for aquestalk.js TTS synthesis - Audio playback via sounddevice - Message queue for sequential TTS playback
This commit is contained in:
@@ -0,0 +1,95 @@
|
||||
/** tts_bridge.js — Persistent Node.js bridge for aquestalk.js TTS synthesis.
|
||||
|
||||
Protocol (via stdin/stdout):
|
||||
- On startup, prints "READY" to stdout.
|
||||
- Reads lines from stdin in format: INPUT_PATH|OUTPUT_PATH
|
||||
- Reads kana text from INPUT_PATH, synthesizes WAV, writes to OUTPUT_PATH.
|
||||
- Prints "OK:OUTPUT_PATH" on success, "ERR:message" on failure.
|
||||
- Exits when stdin is closed.
|
||||
|
||||
Usage:
|
||||
node tts_bridge.js --voice f1 --speed 100
|
||||
*/
|
||||
|
||||
import { readFileSync, writeFileSync } from "fs";
|
||||
import { createInterface } from "readline";
|
||||
import { fileURLToPath, pathToFileURL } from "url";
|
||||
import { dirname, join } from "path";
|
||||
|
||||
const __filename = fileURLToPath(import.meta.url);
|
||||
const __dirname = dirname(__filename);
|
||||
|
||||
// Parse CLI arguments
|
||||
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);
|
||||
}
|
||||
}
|
||||
|
||||
// Import aquestalk.js from sibling directory
|
||||
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);
|
||||
}
|
||||
|
||||
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:${line}\n`);
|
||||
return;
|
||||
}
|
||||
|
||||
const inputPath = line.substring(0, sepIdx);
|
||||
const outputPath = line.substring(sepIdx + 1);
|
||||
|
||||
try {
|
||||
const kanaText = readFileSync(inputPath, "utf-8").trim();
|
||||
|
||||
if (!kanaText) {
|
||||
process.stdout.write(`ERR:EMPTY_TEXT\n`);
|
||||
return;
|
||||
}
|
||||
|
||||
const wav = aq.run(kanaText, speed);
|
||||
|
||||
writeFileSync(outputPath, Buffer.from(wav));
|
||||
|
||||
process.stdout.write(`OK:${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);
|
||||
});
|
||||
});
|
||||
Reference in New Issue
Block a user