feat(kokoro-tts): dual-engine ONNX support with HuggingFace model download

plugins/_kokoro_tts/api/resolve_hf_repo.py

@@ -0,0 +1,61 @@
+import json
+import urllib.error
+import urllib.request
+
+from helpers.api import ApiHandler, Request, Response
+
+
+class ResolveHfRepo(ApiHandler):
+    async def process(self, input: dict, request: Request) -> dict | Response:
+        repo = str(input.get("repo", "") or "").strip()
+        if not repo:
+            return {"success": False, "error": "No repo ID provided."}
+
+        try:
+            url = f"https://huggingface.co/api/models/{repo}"
+            with urllib.request.urlopen(url, timeout=15) as resp:
+                data = json.loads(resp.read())
+        except urllib.error.HTTPError as e:
+            if e.code == 404:
+                return {"success": False, "error": f"Repo '{repo}' not found."}
+            return {"success": False, "error": f"HuggingFace API error: {e}"}
+        except Exception as e:
+            return {"success": False, "error": f"Network error: {e}"}
+
+        files = [s["rfilename"] for s in data.get("siblings", [])]
+
+        # Find model: prefer full-precision .onnx
+        onnx_files = [f for f in files if f.endswith(".onnx")]
+        if not onnx_files:
+            return {"success": False, "error": f"No .onnx file found in {repo}."}
+        model_file = next(
+            (
+                f
+                for f in onnx_files
+                if not any(q in f.lower() for q in ("int8", "quantized", "fp16", "q8"))
+            ),
+            onnx_files[0],
+        )
+
+        # Find voices: .npz or .bin (not .onnx)
+        voices_files = [
+            f
+            for f in files
+            if f.endswith(".npz") or (f.endswith(".bin") and not f.endswith(".onnx"))
+        ]
+        if not voices_files:
+            return {
+                "success": False,
+                "error": f"No voices file (.npz or .bin) found in {repo}.",
+            }
+        # Prefer .npz over .bin
+        npz_files = [f for f in voices_files if f.endswith(".npz")]
+        voices_file = npz_files[0] if npz_files else voices_files[0]
+
+        return {
+            "success": True,
+            "repo": repo,
+            "model_file": model_file,
+            "voices_file": voices_file,
+            "all_files": files,
+        }

plugins/_kokoro_tts/api/status.py

@@ -4,28 +4,37 @@
 from plugins._kokoro_tts.helpers import migration, runtime
 
 
+def _pkg_version(name: str) -> tuple[str, str]:
+    try:
+        return importlib.metadata.version(name), ""
+    except Exception as e:
+        return "", str(e)
+
+
 class Status(ApiHandler):
     async def process(self, input: dict, request: Request) -> dict | Response:
         migration.ensure_migrated()
 
-        package_version = ""
-        package_error = ""
-        try:
-            package_version = importlib.metadata.version("kokoro")
-        except Exception as e:
-            package_error = str(e)
+        cfg = runtime.get_config()
+        py_version, py_error = _pkg_version("kokoro")
+        onnx_version, onnx_error = _pkg_version("kokoro_onnx")
 
         return {
             "plugin": "_kokoro_tts",
             "enabled": runtime.is_globally_enabled(),
-            "config": runtime.get_config(),
+            "config": cfg,
+            "engine": cfg.get("engine", "kokoro_py"),
             "model": {
                 "ready": await runtime.is_downloaded(),
                 "loading": await runtime.is_downloading(),
             },
             "package": {
-                "version": package_version,
-                "error": package_error,
+                "version": py_version,
+                "error": py_error,
+            },
+            "onnx_package": {
+                "version": onnx_version,
+                "error": onnx_error,
             },
             "fallback": "Browser-native speechSynthesis remains the fallback when Kokoro is disabled.",
         }

plugins/_kokoro_tts/default_config.yaml

@@ -1,3 +1,10 @@
 voice: am_puck,am_onyx
 voice_weights: {}
 speed: 1.1
+engine: kokoro_py  # kokoro_py | kokoro_onnx
+lang: en-us  # espeak-ng language code: en-us, de, fr, es, it, etc.
+onnx_hf_repo: ''  # HuggingFace repo ID, e.g. Godelaune/Kokoro-82M-ONNX-German-Martin
+onnx_model_file: ''  # auto-detected from repo if empty; e.g. kokoro-martin.onnx
+onnx_voices_file: ''  # auto-detected from repo if empty; e.g. voices-martin.npz
+onnx_mixed_lang: en-us  # secondary espeak-ng language for matched terms
+onnx_mixed_lang_terms: ''  # comma/newline-separated terms to phonemize with secondary language