feat: add kokoro-tts CLI tool for text-to-speech streaming

kokoro.ts

@@ -0,0 +1,66 @@
+/* Step before building:
+ * 1. Copy onnxruntime-node/bin to out/
+ * 2. Modify onnxruntime-node/dist/binding.js for bun
+ * 3. Profit */
+import os from "node:os";
+import { $ } from "bun";
+import chalk from "chalk";
+import { KokoroTTS, TextSplitterStream } from "kokoro-js";
+
+const model_id = "onnx-community/Kokoro-82M-v1.0-ONNX";
+const tts = await KokoroTTS.from_pretrained(model_id, {
+  dtype: "fp32", // Options: "fp32", "fp16", "q8", "q4", "q4f16"
+  device: "cpu", // Options: "wasm", "webgpu" (web) or "cpu" (node). If using "webgpu", we recommend using dtype="fp32".
+});
+
+const stream = new TextSplitterStream();
+const input = await $`wl-paste`.text();
+if (!input) {
+  console.error("No input");
+  process.exit(1);
+}
+stream.push(input);
+
+await $`rm -r ${os.homedir()}/.aitts/`.nothrow();
+await $`mkdir ${os.homedir()}/.aitts/`.nothrow();
+
+let fileId = 0;
+let buffer = "";
+for (const text of stream) {
+  buffer += ` ${text}`;
+  if (buffer.length < 100) continue;
+
+  console.log(`${chalk.blue(fileId)}:${buffer}`);
+  const audio = await tts.generate(buffer, {
+    voice: "af_heart",
+    speed: 1.2,
+  });
+
+  const file = `${os.homedir()}/.aitts/${fileId}.wav`;
+  audio.save(file);
+
+  fileId++;
+  buffer = "";
+}
+
+if (buffer) {
+  console.log(`${chalk.blue(fileId)}:${buffer}`);
+  const audio = await tts.generate(buffer, {
+    voice: "af_heart",
+    speed: 1.2,
+  });
+
+  const file = `${os.homedir()}/.aitts/${fileId}.wav`;
+  audio.save(file);
+}
+
+console.log(chalk.yellow("Combine wav file with sox..."));
+await $`sox ${os.homedir()}/.aitts/* ${os.homedir()}/.aitts/all.wav`;
+console.log(chalk.green("Success: ~/.aitts/all.wav"));
+
+console.log(chalk.yellow("Listening with mpv..."));
+await Bun.spawn({
+  cmd: ["mpv", `${os.homedir()}/.aitts/all.wav`],
+  stdout: "inherit",
+}).exited;
+console.log(chalk.green("Have a nice day!"));