CUDA 11.8 CUDNN 8

Tested and working

Dockerfile

@@ -1,89 +1,104 @@
 ##########################################
-FROM nvidia/cuda:12.3.2-cudnn9-runtime-ubuntu22.04 AS build
+FROM nvidia/cuda:11.8.0-cudnn8-runtime-ubuntu22.04 AS build
 
 ARG TARGETARCH=linux_x86_64
 ARG WYOMING_PIPER_VERSION="1.5.2"
-ARG ONNXRUNTIME_VERSION="1.20.1"
+ARG PIPER_VERSION="1.2.0"
+ARG PIPER_PHONEMIZE_VERSION="1.1.0"
 
-ENV LANG C.UTF-8
+ENV LANG=C.UTF-8
 ENV DEBIAN_FRONTEND=noninteractive
 
 RUN \
-    apt-get update && apt-get upgrade -y &&\
+    apt-get update &&\
     apt-get install -y --no-install-recommends \
         wget \
         curl \
         vim \
-        git \
         patch \
         python3 \
         python3-dev \
         python3-venv \
         python3-pip \
+        ca-certificates \
+        git \
         build-essential \
         cmake \
-        ca-certificates \
         pkg-config
 
-RUN \
-    mkdir -p /app/lib &&\
-    cd /tmp && wget -q https://github.com/microsoft/onnxruntime/releases/download/v${ONNXRUNTIME_VERSION}/onnxruntime-linux-x64-gpu-${ONNXRUNTIME_VERSION}.tgz &&\
-    tar xzf onnxruntime-linux-x64-gpu-${ONNXRUNTIME_VERSION}.tgz &&\
-    cp -rfv /tmp/onnxruntime-linux-x64-gpu-${ONNXRUNTIME_VERSION}/lib/lib* /app/lib &&\
-    cp -rfv /tmp/onnxruntime-linux-x64-gpu-${ONNXRUNTIME_VERSION}/include/* /usr/local/include/
-
-WORKDIR /build
-
-COPY patches/* /tmp/
-RUN \
-    git clone https://github.com/rhasspy/piper.git /build &&\
-    cd /build && patch -p0 --forward < /tmp/piper_CMakeLists.patch
-
-RUN ONNXRUNTIME_VERSION=${ONNXRUNTIME_VERSION} cmake -Bbuild -DCMAKE_INSTALL_PREFIX=install
-RUN cmake --build build --config Release
-RUN cmake --install build
-
 WORKDIR /app
 
+COPY run.sh .
+COPY patches/* /tmp/
+
 RUN \
+    mkdir -p /app/lib /app/share /app/include/piper-phonemize &&\
     python3 -m venv /app &&\
-    mkdir /app/piper && cp -rfv /build/install/* /app/piper/ &&\
-    chmod 755 /app/piper/piper /app/piper/espeak-ng
-
-RUN /app/piper/piper --help
-
-COPY requirements.txt /app/
-COPY run.sh /app/
-
-RUN \
     . /app/bin/activate && \
-    /app/bin/python3 -m pip install --no-cache-dir \
-        -r /app/requirements.txt \
-        &&\
+    \
+    /app/bin/python3 -m pip install --no-cache-dir --force-reinstall --no-deps \
+        "piper-tts==${PIPER_VERSION}" &&\
+    \
+    wget -q https://github.com/rhasspy/piper-phonemize/releases/download/v${PIPER_PHONEMIZE_VERSION}/piper_phonemize-${PIPER_PHONEMIZE_VERSION}-cp310-cp310-manylinux_2_28_x86_64.whl \
+        -O /tmp/piper_phonemize-${PIPER_PHONEMIZE_VERSION}-py3-none-any.whl &&\
+    \
+    /app/bin/python3 -m pip install --no-cache-dir --force-reinstall --no-deps \
+        /tmp/piper_phonemize-${PIPER_PHONEMIZE_VERSION}-py3-none-any.whl &&\
+    \
+    wget -q https://github.com/rhasspy/piper-phonemize/releases/download/v${PIPER_PHONEMIZE_VERSION}/libpiper_phonemize-amd64.tar.gz -O -| \
+    tar -zxvf - -C /app &&\
+    \
+    mv /app/etc/* /app/share/ &&\
+    mv /app/lib/espeak-ng-data /app/share/ &&\
+    mv /app/include/*.hpp /app/include/piper-phonemize/ &&\
+    mv /app/include/cpu_provider_factory.h /app/include/piper-phonemize/ &&\
+    mv /app/include/provider_options.h /app/include/piper-phonemize/ &&\
+    \
+    LATEST_PIPER_VERSION=$(wget -q "https://api.github.com/repos/rhasspy/piper/releases/latest" -O -|awk '/tag_name/{print $4;exit}' FS='[""]') && \
+    \
+    wget "https://github.com/rhasspy/piper/releases/download/${LATEST_PIPER_VERSION}/piper_${TARGETARCH}.tar.gz" -O -|tar -zxvf - -C /usr/share &&\
     \
     /app/bin/python3 -m pip install --no-cache-dir \
         "wyoming-piper @ https://github.com/rhasspy/wyoming-piper/archive/refs/tags/v${WYOMING_PIPER_VERSION}.tar.gz"
 
+
+WORKDIR /work
+
+RUN git clone https://github.com/rhasspy/piper.git .
+
+RUN cmake -Bbuild -DCMAKE_INSTALL_PREFIX=install -DPIPER_PHONEMIZE_DIR=/app
+RUN cmake --build build --config Release
+RUN cmake --install build
+
+RUN /work/install/piper --help
+
+RUN mkdir -p /app/piper && \
+    mv /work/install/lib*.so* /app/lib/ &&\
+    cp -rf /work/install/* /app/piper/
+
+RUN \
+    wget -q https://github.com/microsoft/onnxruntime/releases/download/v1.14.1/onnxruntime-linux-x64-gpu-1.14.1.tgz -O - | \
+    tar -zxvf - -C /tmp/ &&\
+    cp /tmp/onnxruntime-linux-x64-gpu-1.14.1/lib/* /app/lib/
+
 RUN \
     cd /app/lib/python3.10/site-packages/wyoming_piper/; \
-    for file in /tmp/wyoming_piper*.diff;do patch -p0 --forward < $file;done; \
-    cp /build/install/lib*.so* /app/lib/
+    for file in /tmp/wyoming_piper*.diff;do patch -p0 --forward < $file;done;
 
 ##########################################
-FROM nvidia/cuda:12.3.2-cudnn9-runtime-ubuntu22.04 AS dist
+FROM nvidia/cuda:11.8.0-cudnn8-runtime-ubuntu22.04 AS dist
 
 ENV PYTHONUNBUFFERED=1
+ENV PATH="/bin:$PATH"
 
 RUN \
     mkdir -p /data /app &&\
-    apt-get update && apt-get upgrade -y &&\
+    apt-get update &&\
     apt-get install -y --no-install-recommends \
         wget \
         curl \
         vim \
         python3 \
-        python3-dev \
-        python3-venv \
         python3-pip \
         ca-certificates
 
@@ -92,8 +107,6 @@ RUN rm -rf /root/.cache/pip /var/lib/apt/lists/*
 WORKDIR /app
 COPY --from=build /app .
 
-RUN ln -s /app/piper/espeak-ng-data /usr/share/espeak-ng-data
-
 EXPOSE 10200
 
 ENTRYPOINT ["bash", "/app/run.sh"]

README.md

@@ -9,11 +9,17 @@ https://github.com/rhasspy/wyoming-piper
 
 docker pull ghcr.io/slackr31337/wyoming-piper-gpu:latest
 
-Use environment variable to set piper voice
 
-PIPER_VOICE="en_US-lessac-medium"
-PIPER_LENGTH="1.0"
-PIPER_NOISE="0.667"
-PIPER_NOISEW="0.333"
-PIPER_SPEAKER="0"
-PIPER_SILENCE="1.2"
+# Use environment variable to set piper voice
+
+> PIPER_VOICE="en_US-lessac-medium"
+> 
+> PIPER_LENGTH="1.0"
+> 
+> PIPER_NOISE="0.667"
+> 
+> PIPER_NOISEW="0.333"
+> 
+> PIPER_SPEAKER="0"
+> 
+> PIPER_SILENCE="1.2"

checkCUDA.py

@@ -0,0 +1,94 @@
+"""
+    From https://github.com/rhasspy/piper/issues/24
+"""
+
+import torch
+import os
+
+def print_cuda_details():
+
+    if torch.cuda.is_available():
+        # GPU device details
+        device_id = torch.cuda.current_device()
+        device_name = torch.cuda.get_device_name(device_id)
+        total_memory = torch.cuda.get_device_properties(device_id).total_memory / (1024**3)  # GB
+        memory_allocated = torch.cuda.memory_allocated(device_id) / (1024**3)  # GB
+        memory_cached = torch.cuda.memory_reserved(device_id) / (1024**3)  # GB
+        multiprocessors = torch.cuda.get_device_properties(device_id).multi_processor_count
+        
+        # CUDA and PyTorch versions
+        cuda_version = torch.version.cuda
+        pytorch_version = torch.__version__
+
+        # Device properties
+        device_properties = torch.cuda.get_device_properties(device_id)
+        compute_capability = device_properties.major, device_properties.minor
+        
+        # Display the details with improved formatting
+        print("~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
+        print(f"")
+        print("Checking CUDA information within the python environment...")
+        print(f"")
+
+        print("###########################################################")
+        print(f"CUDA Available:         TRUE")
+        print(f"Using GPU:              {device_name}")
+        print("###########################################################")
+        print(f"")
+
+        print(f"CUDA Version:           {cuda_version}")
+        print(f"PyTorch Version:        {pytorch_version}")
+        print(f"Device ID:              {device_id}")
+        print(f"Total Memory:           {total_memory:.2f} GB")
+        print(f"Memory Allocated:       {memory_allocated:.2f} GB")
+        print(f"Memory Cached:          {memory_cached:.2f} GB")
+        print(f"Multiprocessors:        {multiprocessors}")
+        print(f"Compute Capability:     {compute_capability[0]}.{compute_capability[1]}")
+        print(f"")
+
+        # Device Name
+        print(f"CUDA Device Name:       {torch.cuda.get_device_name(device_id)}")
+        print(f"")
+
+        # CUDA Path (from environment variables)
+        cuda_path = os.getenv('CUDA_PATH', 'Not set')
+        print(f"CUDA Path:              {cuda_path}")
+        
+        # CUDA Toolkit Directory (check via environment variable if available)
+        cuda_toolkit_dir = os.getenv('CUDA_HOME', 'Not set')
+        print(f"CUDA Toolkit Dir:       {cuda_toolkit_dir}")
+        
+        # Environment Variables
+        print(f"\n")
+        print("=================================================")
+        print("--- Below are OS SYSTEM Environment Variables ---")
+        print(" not just whats available within the python VENV")
+        print("=================================================")
+        print(f"")
+
+        print(f"{'CUDA_HOME:':<25} {os.environ.get('CUDA_HOME', 'Not set')}")
+        print(f"")
+
+        print(f"{'CUDA_VISIBLE_DEVICES:':<25} {os.environ.get('CUDA_VISIBLE_DEVICES', 'Not set')}")
+        print(f"")
+
+        print(f"{'LD_LIBRARY_PATH:':<25} {os.environ.get('LD_LIBRARY_PATH', 'Not set')}")
+        print(f"")
+
+        print(f"{'PATH:':<25} {os.environ.get('PATH', 'Not set')}")
+        print(f"")
+
+        print("~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
+    else:
+        print("##############################################################")
+        print("CUDA is NOT available! You may need to reinstall Pytorch/Torch")
+        print("Falling back to using CPU.")
+        print("##############################################################")
+        print(f"")
+
+        # Provide details for CPU-based configurations
+        print(f"PyTorch Version:        {torch.__version__}")
+        print("~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
+
+# Call the function to print details
+print_cuda_details()

patches/piper_CMakeLists.patch

@@ -1,16 +0,0 @@
---- CMakeLists.txt      2025-01-12 11:49:52.100425655 -0500
-+++ CMakeLists.txt.patched      2025-01-12 11:50:56.033517774 -0500
-@@ -60,11 +60,12 @@
-
- if(NOT DEFINED PIPER_PHONEMIZE_DIR)
-   set(PIPER_PHONEMIZE_DIR "${CMAKE_CURRENT_BINARY_DIR}/pi")
-+  set(ONNXRUNTIME_VERSION "$ENV{ONNXRUNTIME_VERSION}")
-   ExternalProject_Add(
-     piper_phonemize_external
-     PREFIX "${CMAKE_CURRENT_BINARY_DIR}/p"
-     URL "https://github.com/rhasspy/piper-phonemize/archive/refs/heads/master.zip"
--    CMAKE_ARGS -DCMAKE_INSTALL_PREFIX:PATH=${PIPER_PHONEMIZE_DIR}
-+    CMAKE_ARGS -DCMAKE_INSTALL_PREFIX:PATH=${PIPER_PHONEMIZE_DIR} -DONNXRUNTIME_VERSION=${ONNXRUNTIME_VERSION}
-   )
-   add_dependencies(piper piper_phonemize_external)
-   add_dependencies(test_piper piper_phonemize_external)

patches/wyoming_piper_process.diff

@@ -1,9 +1,9 @@
---- process.py  2025-01-12 12:08:32.699567820 -0500
-+++ process.py.patched  2025-01-12 12:28:28.568004323 -0500
+--- process.py	2025-01-12 14:43:19.398599231 -0500
++++ process.patched	2025-01-12 14:44:38.795966270 -0500
 @@ -148,6 +148,15 @@
              if self.args.noise_w:
                  piper_args.extend(["--noise-w", str(self.args.noise_w)])
-
+ 
 +            if self.args.sentence_silence:
 +                piper_args.extend(["--sentence-silence", str(self.args.sentence_silence)])
 +
@@ -15,4 +15,4 @@
 +
              _LOGGER.debug(
                  "Starting piper process: %s args=%s", self.args.piper, piper_args
-             )
+             )

requirements.txt

@@ -1,2 +0,0 @@
-piper-tts>=1.2.0
-onnxruntime-gpu==1.20.1

run.sh

@@ -1,6 +1,6 @@
 #!/usr/bin/env bash
 
-export LD_LIBRARY_PATH=/app/lib/:/usr/local/cuda-12.3/targets/x86_64-linux/lib:$LD_LIBRARY_PATH
+export LD_LIBRARY_PATH=/app/lib/:/usr/local/cuda-11.8/targets/x86_64-linux/lib:$LD_LIBRARY_PATH
 
 # Run wyoming-piper server
 source /app/bin/activate
@@ -16,6 +16,4 @@ source /app/bin/activate
     --max-piper-procs "${PIPER_PROCS:-1}" \
     --data-dir /data \
     --download-dir /data \
-    "$@"
-
-# --use-cuda crashes with onnxruntime 1.20.1
+    --use-cuda "$@"