dsh-line-bot 0.9.6 → 0.10.0

package/tools/mlx_synth.py

@@ -0,0 +1,60 @@
+#!/usr/bin/env python3
+"""One-shot MLX CosyVoice3 synthesis helper for dsh-line-bot.
+
+Reads a JSON job file, synthesizes one WAV, and prints a single
+`RESULT:{...}` line on stdout so the Node side can parse it.
+"""
+import glob
+import json
+import os
+import sys
+import traceback
+
+
+def emit(payload):
+    sys.stdout.write("RESULT:" + json.dumps(payload, ensure_ascii=False) + "\n")
+    sys.stdout.flush()
+
+
+def main():
+    if len(sys.argv) < 2:
+        emit({"ok": False, "error": "missing job file"})
+        return 2
+    with open(sys.argv[1], "r", encoding="utf-8") as handle:
+        job = json.load(handle)
+
+    prefix = job.get("prefix") or "tts"
+    os.makedirs(os.path.dirname(prefix) or ".", exist_ok=True)
+
+    try:
+        from mlx_audio.tts.generate import generate_audio
+
+        kwargs = {
+            "text": job["text"],
+            "model": job.get("model") or "mlx-community/Fun-CosyVoice3-0.5B-2512-8bit",
+            "ref_audio": job.get("refAudio"),
+            "file_prefix": prefix,
+            "audio_format": "wav",
+            "verbose": False,
+            "join_audio": True,
+        }
+        if job.get("instructText"):
+            kwargs["instruct_text"] = job["instructText"]
+        elif job.get("refText"):
+            kwargs["ref_text"] = job["refText"]
+        generate_audio(**kwargs)
+
+        candidates = sorted(glob.glob(prefix + "*.wav"), key=os.path.getmtime)
+        if not candidates:
+            emit({"ok": False, "error": "no output wav produced"})
+            return 1
+        emit({"ok": True, "file": candidates[-1]})
+        return 0
+    except Exception as exc:  # noqa: BLE001
+        traceback.print_exc(file=sys.stderr)
+        emit({"ok": False, "error": str(exc)})
+        return 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())