Re-enable TTS (bark.cpp) Inference

CMakeLists.txt

@@ -189,7 +189,7 @@ endif()
 
 # bark_cpp project
 # Temporarily disabled since version v0.0.9.3
-option(BARK_BUILD "Build bark.cpp" OFF)
+option(BARK_BUILD "Build bark.cpp" ON)
 if(BARK_BUILD)
     # Filter out HIPBLAS and Vulkan options for bark.cpp since it doesn't support them
     set(BARK_CMAKE_OPTIONS ${USER_DEFINED_OPTIONS})
@@ -219,10 +219,26 @@ if(WIN32)
     install(
         DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/stable_diffusion_build/bin/Release/
+        DESTINATION ${SKBUILD_PLATLIB_DIR}/nexa/gguf/lib/stable_diffusion
+        USE_SOURCE_PERMISSIONS
+        FILES_MATCHING
+            PATTERN "*.dll"
+    )
+
+    install(
+        DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/llama_build/bin/Release/
+        DESTINATION ${SKBUILD_PLATLIB_DIR}/nexa/gguf/lib/llama
+        USE_SOURCE_PERMISSIONS
+        FILES_MATCHING
+            PATTERN "*.dll"
+    )
+
+    install(
+        DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/bark_build/bin/Release/
             ${CMAKE_CURRENT_BINARY_DIR}/bark_build/Release/
-        DESTINATION ${SKBUILD_PLATLIB_DIR}/nexa/gguf/lib
+        DESTINATION ${SKBUILD_PLATLIB_DIR}/nexa/gguf/lib/bark.cpp
         USE_SOURCE_PERMISSIONS
         FILES_MATCHING
             PATTERN "*.dll"
@@ -231,10 +247,26 @@ if(WIN32)
     install(
         DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/stable_diffusion_build/bin/Release/
+        DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/nexa/gguf/lib/stable_diffusion
+        USE_SOURCE_PERMISSIONS
+        FILES_MATCHING
+            PATTERN "*.dll"
+    )
+
+    install(
+        DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/llama_build/bin/Release/
+        DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/nexa/gguf/lib/llama
+        USE_SOURCE_PERMISSIONS
+        FILES_MATCHING
+            PATTERN "*.dll"
+    )
+
+    install(
+        DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/bark_build/bin/Release/
             ${CMAKE_CURRENT_BINARY_DIR}/bark_build/Release/
-        DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/nexa/gguf/lib
+        DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/nexa/gguf/lib/bark.cpp
         USE_SOURCE_PERMISSIONS
         FILES_MATCHING
             PATTERN "*.dll"
@@ -243,9 +275,27 @@ else()
     install(
         DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/stable_diffusion_build/bin/
+        DESTINATION ${SKBUILD_PLATLIB_DIR}/nexa/gguf/lib/stable_diffusion
+        USE_SOURCE_PERMISSIONS
+        FILES_MATCHING
+            PATTERN "*.so"
+            PATTERN "*.dylib"
+    )
+
+    install(
+        DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/llama_install/lib/
+        DESTINATION ${SKBUILD_PLATLIB_DIR}/nexa/gguf/lib/llama
+        USE_SOURCE_PERMISSIONS
+        FILES_MATCHING
+            PATTERN "*.so"
+            PATTERN "*.dylib"
+    )
+
+    install(
+        DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/bark_install/lib/
-        DESTINATION ${SKBUILD_PLATLIB_DIR}/nexa/gguf/lib
+        DESTINATION ${SKBUILD_PLATLIB_DIR}/nexa/gguf/lib/bark.cpp
         USE_SOURCE_PERMISSIONS
         FILES_MATCHING
             PATTERN "*.so"
@@ -255,12 +305,30 @@ else()
     install(
         DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/stable_diffusion_build/bin/
+        DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/nexa/gguf/lib/stable_diffusion
+        USE_SOURCE_PERMISSIONS
+        FILES_MATCHING
+            PATTERN "*.so"
+            PATTERN "*.dylib"
+    )
+
+    install(
+        DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/llama_install/lib/
+        DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/nexa/gguf/lib/llama
+        USE_SOURCE_PERMISSIONS
+        FILES_MATCHING
+            PATTERN "*.so"
+            PATTERN "*.dylib"
+    )
+
+    install(
+        DIRECTORY
             ${CMAKE_CURRENT_BINARY_DIR}/bark_install/lib/
-        DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/nexa/gguf/lib
+        DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/nexa/gguf/lib/bark.cpp
         USE_SOURCE_PERMISSIONS
         FILES_MATCHING
             PATTERN "*.so"
             PATTERN "*.dylib"
     )
-endif()
+endif()

nexa/gguf/__init__.py

@@ -2,15 +2,13 @@
 from .nexa_inference_text import NexaTextInference
 from .nexa_inference_vlm import NexaVLMInference
 from .nexa_inference_voice import NexaVoiceInference
-
-# Temporarily disabled since version v0.0.9.3
-# from .nexa_inference_tts import NexaTTSInference
+from .nexa_inference_tts import NexaTTSInference
 
 __all__ = [
     "NexaImageInference",
     "NexaTextInference",
     "NexaVLMInference",
     "NexaVoiceInference",
-    #"NexaTTSInference",
+    "NexaTTSInference",
     "NexaAudioLMInference"
 ]

nexa/gguf/bark/bark_cpp.py

@@ -6,7 +6,19 @@
 
 
 # Load the library
-def _load_shared_library(lib_base_name: str):
+def _load_shared_library(lib_base_name: str, lib_subdir_name: str = ''):
+    """
+    Loads a shared library for bark.cpp using ctypes.
+
+    Args:
+        lib_base_name (str): The base name of the shared library. For example, if your shared library file is named 'libbark.so', 
+                             this parameter should be set to 'bark'. The function automatically handles the 'lib' prefix and 
+                             platform-specific file extensions (e.g., .so, .dll, .dylib).
+        lib_subdir_name (str): The name of the subdirectory where the shared library is located. By default, the function looks 
+                               for libraries in the root directory without recursion. If a non-empty string is
+                               provided, the function will search in '<root_dir>/<lib_subdir_name>/' instead.
+    """
+
     # Determine the file extension based on the platform
     if sys.platform.startswith("linux"):
         lib_ext = ".so"
@@ -19,10 +31,16 @@ def _load_shared_library(lib_base_name: str):
 
     # Construct the paths to the possible shared library names
     _base_path = pathlib.Path(__file__).parent.parent.resolve()
-    _lib_paths = [
-        _base_path / f"lib/lib{lib_base_name}{lib_ext}",
-        _base_path / f"lib/{lib_base_name}{lib_ext}",
-    ]
+    if len(lib_subdir_name) == 0:
+        _lib_paths = [
+            _base_path / f"lib/lib{lib_base_name}{lib_ext}",
+            _base_path / f"lib/{lib_base_name}{lib_ext}",
+        ]
+    else:
+        _lib_paths = [
+            _base_path / "lib" / lib_subdir_name / f"lib{lib_base_name}{lib_ext}",
+            _base_path / "lib" / lib_subdir_name / f"{lib_base_name}{lib_ext}",
+        ] 
 
     if "BARK_CPP_LIB" in os.environ:
         lib_base_name = os.environ["BARK_CPP_LIB"]
@@ -32,7 +50,10 @@ def _load_shared_library(lib_base_name: str):
 
     # Add the library directory to the DLL search path on Windows (if needed)
     if sys.platform == "win32" and sys.version_info >= (3, 8):
-        os.add_dll_directory(str(_base_path))
+        if len(lib_subdir_name) == 0:
+            os.add_dll_directory(str(_base_path / 'lib'))
+        else:
+            os.add_dll_directory(str(_base_path / 'lib' / lib_subdir_name))
 
     # Try to load the shared library, handling potential errors
     for _lib_path in _lib_paths:
@@ -49,9 +70,10 @@ def _load_shared_library(lib_base_name: str):
 
 # Specify the base name of the shared library to load
 _lib_base_name = "bark"
+_lib_subdir_name = "bark.cpp"
 
 # Load the library
-_lib = _load_shared_library(_lib_base_name)
+_lib = _load_shared_library(_lib_base_name, _lib_subdir_name)
 
 
 

nexa/gguf/lib_utils.py

@@ -15,9 +15,11 @@ def is_gpu_available():
     return sentinel_file_exists
 
 # Load the library
-def load_library(lib_base_name: str):
+def load_library(lib_base_name: str, lib_subdir_name: str = ''):
     # Construct the paths to the possible shared library names
     _base_path = pathlib.Path(os.path.abspath(os.path.dirname(__file__))) / "lib"
+    if len(lib_subdir_name) != 0:
+        _base_path = _base_path / lib_subdir_name
     # Searching for the library in the current directory under the name "libllama" (default name
     # for llamacpp) and "llama" (default name for this repo)
     _lib_paths: List[pathlib.Path] = []

nexa/gguf/llama/llama_cpp.py

@@ -33,8 +33,9 @@
 
 # Specify the base name of the shared library to load
 _lib_base_name = "llama"
+_lib_subdir_name = 'llama'
 # Load the library
-_lib = load_library(_lib_base_name)
+_lib = load_library(_lib_base_name, _lib_subdir_name)
 
 ctypes_function = ctypes_function_for_shared_library(_lib)
 

nexa/gguf/sd/stable_diffusion_cpp.py

@@ -24,9 +24,10 @@
 
 # Specify the base name of the shared library to load
 _lib_base_name = "stable-diffusion"
+_lib_subdir_name = 'stable_diffusion'
 
 # Load the library
-_lib = load_library(_lib_base_name)
+_lib = load_library(_lib_base_name, _lib_subdir_name)
 
 # ctypes sane type hint helpers
 #

pyproject.toml

@@ -136,7 +136,7 @@ cmake.args = [
     "-DCMAKE_BUILD_PARALLEL_LEVEL=16",
     "-DSTABLE_DIFFUSION_BUILD=ON",
     "-DLLAMA_BUILD=ON",
-    "-DBARK_BUILD=OFF",
+    "-DBARK_BUILD=ON",
 ]
 
 [tool.scikit-build.metadata.version]

tests/test_image_generation.py

@@ -1,6 +1,6 @@
 from nexa.gguf import NexaImageInference
 from tempfile import TemporaryDirectory
-from .utils import download_model
+from tests.utils import download_model
 
 sd = NexaImageInference(
     model_path="sd1-4",

tests/test_tts_generation.py

@@ -1,24 +1,24 @@
 # Temporarily disabled since version v0.0.9.3
 
-# from nexa.gguf import NexaTTSInference
+from nexa.gguf import NexaTTSInference
 
-# def test_tts_generation():
-#     tts = NexaTTSInference(
-#         model_path="bark-small",
-#         local_path=None,
-#         n_threads=4,
-#         seed=42,
-#         sampling_rate=24000,
-#         verbosity=2
-#     )
+def test_tts_generation():
+    tts = NexaTTSInference(
+        model_path="bark-small",
+        local_path=None,
+        n_threads=4,
+        seed=42,
+        sampling_rate=24000,
+        verbosity=2
+    )
 
-#     # Generate audio from prompt
-#     prompt = "Hello, this is a test of the Bark text to speech system."
-#     audio_data = tts.audio_generation(prompt)
+    # Generate audio from prompt
+    prompt = "Hello, this is a test of the Bark text to speech system."
+    audio_data = tts.audio_generation(prompt)
 
-#     # Save the generated audio
-#     tts._save_audio(audio_data, tts.sampling_rate, "tts_output")
-#     print("TTS generation test completed successfully!")
+    # Save the generated audio
+    tts._save_audio(audio_data, tts.sampling_rate, "tts_output")
+    print("TTS generation test completed successfully!")
 
-# if __name__ == "__main__":
-#     test_tts_generation()
+if __name__ == "__main__":
+    test_tts_generation()