From d89521559aa0bb9d3266ff3967eb0319ba9f74ca Mon Sep 17 00:00:00 2001
From: Jamie Pine <ijamespine@me.com>
Date: Sat, 31 Jan 2026 03:09:49 -0800
Subject: [PATCH] Update LocalProvider to manage model size dynamically and
 clean up build scripts

- Introduced a new attribute `_current_model_size` in `LocalProvider` to store the current model size, allowing for dynamic configuration during generation.
- Updated the `generate` method to use the current model size instead of a hardcoded value.
- Modified the `load_model` method to track the requested model size.
- Removed platform-specific extension handling from the build scripts for both CPU and CUDA providers to streamline the build process.
---
 backend/providers/local.py      | 9 +++++----
 providers/pytorch-cpu/build.py  | 4 ----
 providers/pytorch-cuda/build.py | 4 ----
 3 files changed, 5 insertions(+), 12 deletions(-)

diff --git a/backend/providers/local.py b/backend/providers/local.py
index 1bde7bfe..004e6fc3 100644
--- a/backend/providers/local.py
+++ b/backend/providers/local.py
@@ -25,6 +25,7 @@ class LocalProvider:
         """
         self.base_url = base_url.rstrip('/')
         self.client = httpx.AsyncClient(timeout=300.0)  # 5 minute timeout for generation
+        self._current_model_size = "1.7B"  # Default model size
     
     async def generate(
         self,
@@ -42,7 +43,7 @@ class LocalProvider:
                 "voice_prompt": voice_prompt,
                 "language": language,
                 "seed": seed,
-                "model_size": "1.7B",  # TODO: Make configurable
+                "model_size": self._current_model_size,
             }
         )
         response.raise_for_status()
@@ -116,9 +117,9 @@ class LocalProvider:
     
     async def load_model(self, model_size: str) -> None:
         """Load TTS model."""
-        # Model loading is handled automatically by the provider server
-        # when generate() is called, so this is a no-op
-        pass
+        # Track the requested model size - the provider server will load it
+        # when generate() is called with this size
+        self._current_model_size = model_size
     
     def unload_model(self) -> None:
         """Unload model to free memory."""
diff --git a/providers/pytorch-cpu/build.py b/providers/pytorch-cpu/build.py
index 2276b686..8894edcb 100644
--- a/providers/pytorch-cpu/build.py
+++ b/providers/pytorch-cpu/build.py
@@ -56,10 +56,6 @@ def build_provider():
         '--hidden-import', 'librosa',
     ])
     
-    # Platform-specific extensions
-    if platform.system() == "Windows":
-        args[2] = 'tts-provider-pytorch-cpu.exe'
-    
     args.extend([
         '--noconfirm',
         '--clean',
diff --git a/providers/pytorch-cuda/build.py b/providers/pytorch-cuda/build.py
index c6e56e7b..8ff67334 100644
--- a/providers/pytorch-cuda/build.py
+++ b/providers/pytorch-cuda/build.py
@@ -58,10 +58,6 @@ def build_provider():
         '--hidden-import', 'librosa',
     ])
     
-    # Platform-specific extensions
-    if platform.system() == "Windows":
-        args[2] = 'tts-provider-pytorch-cuda.exe'
-    
     args.extend([
         '--noconfirm',
         '--clean',