In this directory, you will find examples on how you could use IPEX-LLM optimize_model
API to accelerate SpeechT5 models. For illustration purposes, we utilize the microsoft/speecht5_tts as reference SpeechT5 models.
To run these examples with IPEX-LLM, we have some recommended requirements for your machine, please refer to here for more information.
In the example synthesize_speech.py, we show a basic use case for SpeechT5 model to synthesize speech based on the given text, with IPEX-LLM INT4 optimizations.
We suggest using conda to manage environment:
conda create -n llm python=3.11
conda activate llm
# below command will install intel_extension_for_pytorch==2.1.10+xpu as default
pip install --pre --upgrade ipex-llm[xpu] --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
pip install "datasets==2.16.1" soundfile # additional package required for SpeechT5 to conduct generation
We suggest using conda to manage environment:
conda create -n llm python=3.11 libuv
conda activate llm
# below command will install intel_extension_for_pytorch==2.1.10+xpu as default
pip install --pre --upgrade ipex-llm[xpu] --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
pip install "datasets==2.16.1" soundfile # additional package required for SpeechT5 to conduct generation
Note
Skip this step if you are running on Windows.
This is a required step on Linux for APT or offline installed oneAPI. Skip this step for PIP-installed oneAPI.
source /opt/intel/oneapi/setvars.sh
For optimal performance, it is recommended to set several environment variables. Please check out the suggestions based on your device.
For Intel Arc™ A-Series Graphics and Intel Data Center GPU Flex Series
export USE_XETLA=OFF
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
export SYCL_CACHE_PERSISTENT=1
For Intel Data Center GPU Max Series
export LD_PRELOAD=${LD_PRELOAD}:${CONDA_PREFIX}/lib/libtcmalloc.so
export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
export SYCL_CACHE_PERSISTENT=1
export ENABLE_SDP_FUSION=1
Note: Please note that
libtcmalloc.so
can be installed byconda install -c conda-forge -y gperftools=2.10
.
For Intel iGPU
export SYCL_CACHE_PERSISTENT=1
For Intel iGPU and Intel Arc™ A-Series Graphics
set SYCL_CACHE_PERSISTENT=1
Note
For the first time that each model runs on Intel iGPU/Intel Arc™ A300-Series or Pro A60, it may take several minutes to compile.
python ./synthesize_speech.py --text 'Artificial intelligence refers to the development of computer systems that can perform tasks that typically require human intelligence.'
In the example, several arguments can be passed to satisfy your requirements:
--repo-id-or-model-path REPO_ID_OR_MODEL_PATH
: argument defining the huggingface repo id for the SpeechT5 model (e.gmicrosoft/speecht5_tts
) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be'microsoft/speecht5_tts'
.--repo-id-or-vocoder-path REPO_ID_OR_VOCODER_PATH
: argument defining the huggingface repo id for the SpeechT5 vocoder (e.gmicrosoft/speecht5_hifigan
, which generates audio from a spectrogram) to be downloaded, or the path to the huggingface checkpoint folder. It is default to be'microsoft/speecht5_hifigan'
.--repo-id-or-data-path REPO_ID_OR_DATA_PATH
: argument defining the huggingface repo id for the audio dataset (e.g.Matthijs/cmu-arctic-xvectors
, which decides voice characteristics) to be downloaded, or the path to the huggingface dataset folder. It is default to be'Matthijs/cmu-arctic-xvectors'
.--text TEXT
: argument defining the text to synthesize speech. It is default to be"Artificial intelligence refers to the development of computer systems that can perform tasks that typically require human intelligence."
.
Text: Artificial intelligence refers to the development of computer systems that can perform tasks that typically require human intelligence.