diff --git a/api/src/core/config.py b/api/src/core/config.py
index 29ab155e..b7d69ca6 100644
--- a/api/src/core/config.py
+++ b/api/src/core/config.py
@@ -35,7 +35,7 @@ class Settings(BaseSettings):
     )
     use_gpu: bool = True  # Whether to use GPU acceleration if available
     device_type: str | None = (
-        None  # Will be auto-detected if None, can be "cuda", "mps", or "cpu"
+        None  # Will be auto-detected if None, can be "cuda", "mps", "xpu", or "cpu"
     )
     allow_local_voice_saving: bool = (
         False  # Whether to allow saving combined voices locally
@@ -75,7 +75,7 @@ class Settings(BaseSettings):
     cors_enabled: bool = True  # Whether to enable CORS
 
     # Temp File Settings for WEB Ui
-    temp_file_dir: str = "api/temp_files"  # Directory for temporary audio files (relative to project root)
+    temp_file_dir: str = "/tmp/kokoro_temp"  # Directory for temporary audio files
     max_temp_dir_size_mb: int = 2048  # Maximum size of temp directory (2GB)
     max_temp_dir_age_hours: int = 1  # Remove temp files older than 1 hour
     max_temp_dir_count: int = 3  # Maximum number of temp files to keep
@@ -96,6 +96,14 @@ def get_device(self) -> str:
             return "mps"
         elif torch.cuda.is_available():
             return "cuda"
+
+        # Check for Intel GPU (XPU)
+        try:
+            if hasattr(torch, "xpu") and torch.xpu.is_available():
+                return "xpu"
+        except Exception:
+            pass
+            
         return "cpu"
 
 
diff --git a/api/src/inference/kokoro_v1.py b/api/src/inference/kokoro_v1.py
index dc11ae3a..0ee25513 100644
--- a/api/src/inference/kokoro_v1.py
+++ b/api/src/inference/kokoro_v1.py
@@ -75,6 +75,8 @@ async def load_model(self, path: str) -> None:
                 self._model = self._model.to(torch.device("mps"))
             elif self._device == "cuda":
                 self._model = self._model.cuda()
+            elif self._device == "xpu":
+                self._model = self._model.to("xpu")
             else:
                 self._model = self._model.cpu()
 
@@ -350,6 +352,12 @@ def _check_memory(self) -> bool:
         if self._device == "cuda":
             memory_gb = torch.cuda.memory_allocated() / 1e9
             return memory_gb > model_config.pytorch_gpu.memory_threshold
+        elif self._device == "xpu":
+            try:
+                memory_gb = torch.xpu.memory_allocated() / 1e9
+                return memory_gb > model_config.pytorch_gpu.memory_threshold
+            except (RuntimeError, AttributeError):
+                pass
         # MPS doesn't provide memory management APIs
         return False
 
@@ -362,6 +370,12 @@ def _clear_memory(self) -> None:
             # Empty cache if available (future-proofing)
             if hasattr(torch.mps, "empty_cache"):
                 torch.mps.empty_cache()
+        elif self._device == "xpu":
+            try:
+                torch.xpu.empty_cache()
+                torch.xpu.synchronize()
+            except (RuntimeError, AttributeError):
+                pass
 
     def unload(self) -> None:
         """Unload model and free resources."""
@@ -375,6 +389,12 @@ def unload(self) -> None:
         if torch.cuda.is_available():
             torch.cuda.empty_cache()
             torch.cuda.synchronize()
+        try:
+            if hasattr(torch, "xpu") and torch.xpu.is_available():
+                torch.xpu.empty_cache()
+                torch.xpu.synchronize()
+        except Exception:
+            pass
 
     @property
     def is_loaded(self) -> bool:
diff --git a/api/src/inference/model_manager.py b/api/src/inference/model_manager.py
index 231f1f97..1cf5533e 100644
--- a/api/src/inference/model_manager.py
+++ b/api/src/inference/model_manager.py
@@ -32,7 +32,7 @@ def __init__(self, config: Optional[ModelConfig] = None):
 
     def _determine_device(self) -> str:
         """Determine device based on settings."""
-        return "cuda" if settings.use_gpu else "cpu"
+        return settings.get_device()
 
     async def initialize(self) -> None:
         """Initialize Kokoro V1 backend."""
diff --git a/api/src/main.py b/api/src/main.py
index 11bd3e5b..3437b006 100644
--- a/api/src/main.py
+++ b/api/src/main.py
@@ -94,6 +94,8 @@ async def lifespan(app: FastAPI):
         startup_msg += "\nUsing Apple Metal Performance Shaders (MPS)"
     elif device == "cuda":
         startup_msg += f"\nCUDA: {torch.cuda.is_available()}"
+    elif device == "xpu":
+        startup_msg += f"\nXPU: {torch.xpu.is_available()}"
     else:
         startup_msg += "\nRunning on CPU"
     startup_msg += f"\n{voicepack_count} voice packs loaded"
diff --git a/docker-bake.hcl b/docker-bake.hcl
index 70e49ba8..2cc0b52d 100644
--- a/docker-bake.hcl
+++ b/docker-bake.hcl
@@ -115,6 +115,21 @@ target "_rocm_base" {
     ]
 }
 
+# Base settings for Intel builds
+target "_intel_base" {
+    inherits = ["_common"]
+    dockerfile = "docker/intel/Dockerfile"
+    labels = {
+        "org.opencontainers.image.title"       = "Kokoro-FastAPI (Intel)"
+        "org.opencontainers.image.description" = "Kokoro TTS served via FastAPI. Intel GPU build (amd64 only)."
+    }
+    annotations = [
+        "org.opencontainers.image.title=Kokoro-FastAPI (Intel)",
+        "org.opencontainers.image.description=Kokoro TTS served via FastAPI. Intel GPU build (amd64 only).",
+    ]
+}
+
+
 
 # Individual platform targets for debugging/testing
 target "cpu-amd64" {
@@ -183,6 +198,15 @@ target "rocm-amd64" {
     ]
 }
 
+# Intel only supports x86
+target "intel-amd64" {
+    inherits = ["_intel_base"]
+    platforms = ["linux/amd64"]
+    tags = [
+        "${REGISTRY}/${OWNER}/${REPO}-intel:${VERSION}-amd64"
+    ]
+}
+
 # Development targets for faster local builds
 target "cpu-dev" {
     inherits = ["_cpu_base"]
@@ -223,10 +247,14 @@ group "rocm-all" {
     targets = ["rocm-amd64"]
 }
 
+group "intel-all" {
+    targets = ["intel-amd64"]
+}
+
 group "all" {
-    targets = ["cpu", "gpu-amd64", "gpu-arm64", "gpu-cu128-amd64", "rocm-amd64"]
+    targets = ["cpu", "gpu-amd64", "gpu-arm64", "gpu-cu128-amd64", "rocm-amd64", "intel-amd64"]
 }
 
 group "individual-platforms" {
-    targets = ["cpu-amd64", "cpu-arm64", "gpu-amd64", "gpu-arm64", "gpu-cu128-amd64", "rocm-amd64"]
+    targets = ["cpu-amd64", "cpu-arm64", "gpu-amd64", "gpu-arm64", "gpu-cu128-amd64", "rocm-amd64", "intel-amd64"]
 }
diff --git a/docker/intel/Dockerfile b/docker/intel/Dockerfile
new file mode 100644
index 00000000..4c8af019
--- /dev/null
+++ b/docker/intel/Dockerfile
@@ -0,0 +1,104 @@
+# Stage 1: Builder - Use devel image for compilation
+FROM ubuntu:24.04 AS builder
+
+# Install Python and build dependencies
+RUN apt-get update -y && \
+    apt-get install -y python3.10 python3-venv python3-dev git curl wget gnupg2 && \
+    apt-get clean && rm -rf /var/lib/apt/lists/* && \
+    curl -LsSf https://astral.sh/uv/install.sh | sh && \
+    mv /root/.local/bin/uv /usr/local/bin/ && \
+    mv /root/.local/bin/uvx /usr/local/bin/
+
+WORKDIR /app
+
+# Copy dependency files
+COPY pyproject.toml ./pyproject.toml
+
+# Install dependencies with intel extras
+ENV UV_HTTP_TIMEOUT=120 UV_HTTP_RETRIES=3 \
+    UV_PYTHON_INSTALL_DIR=/opt/uv-python
+
+RUN uv venv --python 3.10 && \
+    uv sync --extra intel --no-cache --no-install-project
+
+# Stage 2: Runtime
+FROM ubuntu:24.04
+
+# Install runtime dependencies + uv
+RUN apt-get update -y && \
+    apt-get install -y python3.10 espeak-ng espeak-ng-data libsndfile1 ffmpeg curl wget gnupg2 libnuma1 ocl-icd-libopencl1 libze1 intel-media-va-driver-non-free libmfx1 && \
+    curl -LsSf https://astral.sh/uv/install.sh | sh && \
+    mv /root/.local/bin/uv /usr/local/bin/ && \
+    mv /root/.local/bin/uvx /usr/local/bin/ && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+# Add Intel oneAPI repository for MKL and other tools
+RUN wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | gpg --dearmor | tee /usr/share/keyrings/oneapi-archive-keyring.gpg > /dev/null && \
+    echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg] https://apt.repos.intel.com/oneapi all main" | tee /etc/apt/sources.list.d/oneAPI.list && \
+    apt-get update && \
+    apt-get install -y intel-oneapi-mkl intel-oneapi-runtime-mkl intel-oneapi-runtime-compilers intel-oneapi-compiler-dpcpp-cpp-runtime-2025.3 && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+# Install Intel GPU drivers (Compute Runtime & IGC) using the reference script
+COPY docker/intel/install_ubuntu_gpu_drivers.sh /tmp/install_gpu_drivers.sh
+ARG INSTALL_DRIVER_VERSION="25.35.35096"
+RUN chmod +x /tmp/install_gpu_drivers.sh && \
+    /tmp/install_gpu_drivers.sh && \
+    rm -f /tmp/install_gpu_drivers.sh
+
+# Setup user and GID
+ARG RENDER_GID=992
+RUN mkdir -p /usr/share/espeak-ng-data && \
+    ln -s /usr/lib/*/espeak-ng-data/* /usr/share/espeak-ng-data/ && \
+    (groupmod -g ${RENDER_GID} render || groupadd -g ${RENDER_GID} render) && \
+    useradd -m -u 1001 -G video,render appuser && \
+    mkdir -p /app/api/src/models/v1_0 && \
+    chown -R appuser:appuser /app
+
+WORKDIR /app
+
+# Copy uv-managed Python interpreter (the venv's bin/python symlinks into here)
+COPY --from=builder /opt/uv-python /opt/uv-python
+
+# Copy virtual environment from builder
+COPY --from=builder --chown=appuser:appuser /app/.venv /app/.venv
+
+# Download model in runtime stage so download_model.py is present for runtime re-downloads via entrypoint
+COPY --chown=appuser:appuser docker/scripts/download_model.py ./download_model.py
+ARG DOWNLOAD_MODEL=true
+RUN if [ "$DOWNLOAD_MODEL" = "true" ]; then \
+    /app/.venv/bin/python download_model.py --output api/src/models/v1_0 && \
+    chown -R appuser:appuser /app/api/src/models; \
+    fi
+
+# Japanese support requires the UniDic dictionary
+ARG INCLUDE_JAPANESE=true
+RUN if [ "$INCLUDE_JAPANESE" = "true" ]; then \
+    /app/.venv/bin/python -m unidic download && \
+    chown -R appuser:appuser /app/.venv/lib/python*/site-packages/unidic; \
+    fi
+
+# Copy project files
+COPY --chown=appuser:appuser pyproject.toml ./pyproject.toml
+COPY --chown=appuser:appuser api ./api
+COPY --chown=appuser:appuser web ./web
+COPY --chown=appuser:appuser VERSION ./VERSION
+COPY --chown=appuser:appuser docker/intel/entrypoint.sh ./entrypoint.sh
+RUN chmod +x ./entrypoint.sh
+
+USER appuser
+
+# Set environment variables
+ENV PATH="/app/.venv/bin:$PATH" \
+    PYTHONUNBUFFERED=1 \
+    PYTHONPATH=/app:/app/api \
+    UV_LINK_MODE=copy \
+    USE_GPU=true \
+    PHONEMIZER_ESPEAK_PATH=/usr/bin \
+    PHONEMIZER_ESPEAK_DATA=/usr/share/espeak-ng-data \
+    ESPEAK_DATA_PATH=/usr/share/espeak-ng-data \
+    DEVICE="intel" \
+    LD_LIBRARY_PATH="/opt/intel/oneapi/mkl/latest/lib:/opt/intel/oneapi/mkl/latest/lib/intel64:/opt/intel/oneapi/compiler/latest/lib/intel64:$LD_LIBRARY_PATH"
+
+# Run FastAPI server
+CMD ["./entrypoint.sh"]
diff --git a/docker/intel/README.md b/docker/intel/README.md
new file mode 100644
index 00000000..95d11576
--- /dev/null
+++ b/docker/intel/README.md
@@ -0,0 +1,43 @@
+# Kokoro-FastAPI: Intel GPU (XPU) Setup
+
+This directory contains the configuration to run Kokoro-FastAPI leveraging Intel GPUs (Arc, Data Center, and Integrated Graphics) through the Intel Extension for PyTorch (IPEX).
+
+## Features
+- **Intel XPU Acceleration**: Uses IPEX 2.5.10 + PyTorch 2.5.1 for optimized inference.
+- **Automated Driver Stack**: Integrated with Intel's reference driver installation script.
+- **Secure Non-Root User**: Runs as a standard `appuser` with correct GPU group permissions.
+
+## Requirements
+- **Host Drivers**: Ensure your host has the Intel GPU drivers installed (Compute Runtime, Level Zero).
+- **Docker**: Version 20.10+ with `device_cgroup_rules` support.
+
+## Usage
+
+### 1. Build and Start
+Run the following command from the project root:
+```bash
+docker compose -f docker/intel/docker-compose.yml up --build
+```
+
+### 2. Verify GPU Access
+Check the container logs for XPU registration messages or run:
+```bash
+docker compose -f docker/intel/docker-compose.yml exec kokoro-tts python -c "import torch; import intel_extension_for_pytorch; print(f'XPU Available: {torch.xpu.is_available()}')"
+```
+
+## Technical Details
+- **Base Image**: Ubuntu 24.04 (required for glibc compatibility).
+- **Versions**:
+  - Python: 3.12
+  - Torch: 2.5.1
+  - IPEX: 2.5.10+xpu
+- **Groups**: The container uses a default `RENDER_GID=992`. If your host's `render` group has a different ID, pass it as a build argument:
+  ```bash
+  docker compose -f docker/intel/docker-compose.yml build --build-arg RENDER_GID=$(stat -c '%g' /dev/dri/renderD128)
+  ```
+
+## Troubleshooting
+If you see `RuntimeError: Native API failed`, check:
+1. That `/dev/dri` is correctly mapped.
+2. That your host user has permissions to access `/dev/dri` (usually by being in the `video` or `render` groups).
+3. That the `RENDER_GID` build argument matches your host's GID.
diff --git a/docker/intel/docker-compose.yml b/docker/intel/docker-compose.yml
new file mode 100644
index 00000000..daf47e75
--- /dev/null
+++ b/docker/intel/docker-compose.yml
@@ -0,0 +1,28 @@
+name: kokoro-tts-intel
+services:
+  kokoro-tts:
+    image: kokoro-tts-intel
+    build:
+      context: ../..
+      dockerfile: docker/intel/Dockerfile
+      args:
+        RENDER_GID: ${RENDER_GID:-992}
+    volumes:
+      - ../../api:/app/api
+      - models_data:/app/api/src/models/v1_0
+    user: appuser
+    ports:
+      - "8880:8880"
+    devices:
+      - /dev/dri:/dev/dri # Pass Intel GPU device
+    environment:
+      - PYTHONPATH=/app:/app/api
+      - USE_GPU=true
+      - PYTHONUNBUFFERED=1
+      - API_LOG_LEVEL=DEBUG
+      - DEVICE_TYPE=xpu
+    device_cgroup_rules:
+      - 'c 226:* rmw'
+
+volumes:
+  models_data:
diff --git a/docker/intel/entrypoint.sh b/docker/intel/entrypoint.sh
new file mode 100644
index 00000000..179ea53f
--- /dev/null
+++ b/docker/intel/entrypoint.sh
@@ -0,0 +1,14 @@
+#!/bin/bash
+set -e
+
+# In this simplified version, we assume permissions are handled via Dockerfile/Compose
+# and the user is 'appuser'.
+
+if [ "$DOWNLOAD_MODEL" = "true" ]; then
+    echo "Downloading model..."
+    python download_model.py --output api/src/models/v1_0
+fi
+
+echo "Starting Application..."
+# We use 'exec' so the process handles signals
+exec uv run --extra $DEVICE --no-sync python -m uvicorn api.src.main:app --host 0.0.0.0 --port 8880 --log-level debug
diff --git a/docker/intel/install_ubuntu_gpu_drivers.sh b/docker/intel/install_ubuntu_gpu_drivers.sh
new file mode 100755
index 00000000..e02aaf6f
--- /dev/null
+++ b/docker/intel/install_ubuntu_gpu_drivers.sh
@@ -0,0 +1,99 @@
+#!/bin/bash
+#
+# Copyright (c) 2021 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+set -e
+
+if [ -z "$INSTALL_DRIVER_VERSION" ]; then
+    echo "Error: INSTALL_DRIVER_VERSION cannot be empty."
+    exit 1
+fi
+
+apt-get update && apt-get install -y libnuma1 ocl-icd-libopencl1 --no-install-recommends && rm -rf /var/lib/apt/lists/* && \
+case $INSTALL_DRIVER_VERSION in \
+"24.26.30049") \
+        mkdir /tmp/gpu_deps && cd /tmp/gpu_deps ; \
+        curl -L -O https://github.com/intel/compute-runtime/releases/download/24.26.30049.6/intel-level-zero-gpu_1.3.30049.6_amd64.deb ; \
+        curl -L -O https://github.com/intel/compute-runtime/releases/download/24.26.30049.6/intel-opencl-icd_24.26.30049.6_amd64.deb ; \
+        curl -L -O https://github.com/intel/compute-runtime/releases/download/24.26.30049.6/libigdgmm12_22.3.20_amd64.deb ; \
+        curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/igc-1.0.17193.4/intel-igc-core_1.0.17193.4_amd64.deb ; \
+        curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/igc-1.0.17193.4/intel-igc-opencl_1.0.17193.4_amd64.deb ; \
+        dpkg -i *.deb && rm -Rf /tmp/gpu_deps ; \
+;; \
+"24.39.31294") \
+        mkdir /tmp/gpu_deps && cd /tmp/gpu_deps ; \
+        curl -L -O https://github.com/intel/compute-runtime/releases/download/24.39.31294.12/intel-level-zero-gpu_1.6.31294.12_amd64.deb ; \
+        curl -L -O https://github.com/intel/compute-runtime/releases/download/24.39.31294.12/intel-opencl-icd_24.39.31294.12_amd64.deb ; \
+        curl -L -O https://github.com/intel/compute-runtime/releases/download/24.39.31294.12/libigdgmm12_22.5.2_amd64.deb ; \
+        curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/igc-1.0.17791.9/intel-igc-core_1.0.17791.9_amd64.deb ; \
+        curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/igc-1.0.17791.9/intel-igc-opencl_1.0.17791.9_amd64.deb ; \
+        dpkg -i *.deb && rm -Rf /tmp/gpu_deps ; \
+;; \
+"24.52.32224") \
+        mkdir /tmp/gpu_deps && cd /tmp/gpu_deps ; \
+        curl -L -O https://github.com/intel/compute-runtime/releases/download/24.52.32224.5/intel-level-zero-gpu_1.6.32224.5_amd64.deb ; \
+        curl -L -O https://github.com/intel/compute-runtime/releases/download/24.52.32224.5/intel-opencl-icd_24.52.32224.5_amd64.deb ; \
+        curl -L -O https://github.com/intel/compute-runtime/releases/download/24.52.32224.5/libigdgmm12_22.5.5_amd64.deb ; \
+        curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/v2.5.6/intel-igc-core-2_2.5.6+18417_amd64.deb ; \
+        curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/v2.5.6/intel-igc-opencl-2_2.5.6+18417_amd64.deb ; \
+        dpkg -i *.deb && rm -Rf /tmp/gpu_deps ; \
+;; \
+"25.31.34666") \
+	mkdir /tmp/gpu_deps && cd /tmp/gpu_deps ; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.31.34666.3/libze-intel-gpu1_25.31.34666.3-0_amd64.deb; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.31.34666.3/intel-opencl-icd_25.31.34666.3-0_amd64.deb; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.31.34666.3/libigdgmm12_22.8.1_amd64.deb; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.31.34666.3/intel-ocloc_25.31.34666.3-0_amd64.deb; \
+	curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/v2.16.0/intel-igc-core-2_2.16.0+19683_amd64.deb; \
+	curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/v2.16.0/intel-igc-opencl-2_2.16.0+19683_amd64.deb; \
+	dpkg -i *.deb && rm -Rf /tmp/gpu_deps ; \
+;; \
+"25.35.35096") \
+	mkdir /tmp/gpu_deps && cd /tmp/gpu_deps ; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.35.35096.9/libze-intel-gpu1_25.35.35096.9-0_amd64.deb; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.35.35096.9/intel-opencl-icd_25.35.35096.9-0_amd64.deb; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.35.35096.9/libigdgmm12_22.8.1_amd64.deb; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.35.35096.9/intel-ocloc_25.35.35096.9-0_amd64.deb; \
+	curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/v2.18.5/intel-igc-core-2_2.18.5+19820_amd64.deb; \
+	curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/v2.18.5/intel-igc-opencl-2_2.18.5+19820_amd64.deb; \
+	dpkg -i *.deb && rm -Rf /tmp/gpu_deps ; \
+;; \
+"25.48.36300") \
+	mkdir /tmp/gpu_deps && cd /tmp/gpu_deps ; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.48.36300.8/libze-intel-gpu1_25.48.36300.8-0_amd64.deb; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.48.36300.8/intel-opencl-icd_25.48.36300.8-0_amd64.deb; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.48.36300.8/libigdgmm12_22.8.2_amd64.deb; \
+	curl -L -O https://github.com/intel/compute-runtime/releases/download/25.48.36300.8/intel-ocloc_25.48.36300.8-0_amd64.deb; \
+	curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/v2.24.8/intel-igc-core-2_2.24.8+20344_amd64.deb; \
+	curl -L -O https://github.com/intel/intel-graphics-compiler/releases/download/v2.24.8/intel-igc-opencl-2_2.24.8+20344_amd64.deb; \
+	dpkg -i *.deb && rm -Rf /tmp/gpu_deps ; \
+;; \
+*) \
+        dpkg -P intel-gmmlib intel-igc-core intel-igc-opencl intel-level-zero-gpu intel-ocloc intel-opencl intel-opencl-icd && \
+        apt-get update && apt-get -y --no-install-recommends install dpkg-dev && rm -rf /var/lib/apt/lists/* && \
+        cd /drivers/${INSTALL_DRIVER_VERSION} && \
+            dpkg-scanpackages .  > Packages && \
+            cd - ; \
+        echo "deb [trusted=yes arch=amd64] file:/drivers/${INSTALL_DRIVER_VERSION} ./" > /etc/apt/sources.list.d/intel-graphics-${INSTALL_DRIVER_VERSION}.list ; \
+        apt-get update && \
+        apt-get install -y --no-install-recommends \
+            intel-opencl-icd \
+            intel-level-zero-gpu level-zero \
+            intel-media-va-driver-non-free libmfx1 && \
+            rm -rf /var/lib/apt/lists/* ; \
+esac
+
+apt-get clean && rm -rf /var/lib/apt/lists/* && rm -rf /tmp/*
diff --git a/pyproject.toml b/pyproject.toml
index 48ad1eee..440457fe 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -40,6 +40,8 @@ dependencies = [
     "phonemizer-fork>=3.3.2",
     "av>=14.2.0",
     "text2num>=2.5.1",
+    "sentencepiece>=0.2.0",
+    "protobuf>=5.29.1",
 ]
 
 [project.optional-dependencies]
@@ -59,6 +61,15 @@ rocm = [
     "torch==2.8.0+rocm6.4",
     "pytorch-triton-rocm>=3.2.0",
 ]
+intel = [
+    "torch==2.6.0+xpu",
+    "torchaudio==2.6.0+xpu",
+    "torchvision==0.21.0+xpu",
+    "triton-xpu",
+    "pytorch-triton-xpu",
+    "intel-sycl-rt",
+    "intel-opencl-rt",
+]
 test = [
     "pytest==8.3.5",
     "pytest-cov==6.0.0",
@@ -79,6 +90,7 @@ conflicts = [
         { extra = "gpu" },
         { extra = "gpu-cu128" },
         { extra = "rocm" },
+        { extra = "intel" },
     ],
 ]
 override-dependencies = [
@@ -93,6 +105,25 @@ torch = [
     { index = "pytorch-cu128", extra = "gpu-cu128", marker = "platform_machine == 'x86_64'" },
     { index = "pytorch-cu129", extra = "gpu-cu128", marker = "platform_machine == 'aarch64'" },
     { index = "pytorch-rocm", extra = "rocm" },
+    { index = "pytorch-intel", extra = "intel" },
+]
+torchaudio = [
+    { index = "pytorch-intel", extra = "intel" },
+]
+torchvision = [
+    { index = "pytorch-intel", extra = "intel" },
+]
+triton-xpu = [
+    { index = "pytorch-intel", extra = "intel" },
+]
+pytorch-triton-xpu = [
+    { index = "pytorch-intel", extra = "intel" },
+]
+intel-sycl-rt = [
+    { index = "pytorch-intel", extra = "intel" },
+]
+intel-opencl-rt = [
+    { index = "pytorch-intel", extra = "intel" },
 ]
 pytorch-triton-rocm = [
     { index = "pytorch-rocm", extra = "rocm" },
@@ -123,6 +154,13 @@ name = "pytorch-rocm"
 url = "https://download.pytorch.org/whl/rocm6.4"
 explicit = true
 
+[[tool.uv.index]]
+name = "pytorch-intel"
+url = "https://download.pytorch.org/whl/xpu"
+explicit = true
+
+
+
 [build-system]
 requires = ["setuptools>=61.0"]
 build-backend = "setuptools.build_meta"