diff --git a/README.md b/README.md
index 8086494c06a269371a239c43ef0a10ed3317636e..de2864e758f527f5aa16a91ab1738b79acb9d675 100644
--- a/README.md
+++ b/README.md
@@ -456,6 +456,7 @@ DeepSparkHub甄选上百个应用算法和模型，覆盖AI和通用计算各领
 
 | Model                                                                 | Framework    | Dataset     | IXUCA SDK |
 |-----------------------------------------------------------------------|--------------|-------------|-------|
+|  [CosyVoice2-0.5B](audio/speech_synthesis/cosyvoice/pytorch)            |  DeepSpeed     |  openslr      | 4.3.0 |
 |  [PP-TTS-FastSpeech2](audio/speech_synthesis/fastspeech2/paddlepaddle)  |  PaddlePaddle  |  CSMSC        | 3.1.0 |
 |  [PP-TTS-HiFiGAN](audio/speech_synthesis/hifigan/paddlepaddle)          |  PaddlePaddle  |  CSMSC        | 3.1.0 |
 |  [Tacotron2](audio/speech_synthesis/tacotron2/pytorch)                  |  PyTorch       |  LJSpeech     | 2.2.0 |
diff --git a/README_en.md b/README_en.md
index b31e48663f67f9e00b6404b42e18cd3ea514e035..3116386737b60d693bde9627ba883a1c87b74a59 100644
--- a/README_en.md
+++ b/README_en.md
@@ -458,6 +458,7 @@ individuals, healthcare, education, communication, energy, and more.
 
 | Model                                                                 | Framework    | Dataset     | IXUCA SDK |
 |-----------------------------------------------------------------------|--------------|-------------|-------|
+|  [CosyVoice2-0.5B](audio/speech_synthesis/cosyvoice/pytorch)            |  DeepSpeed     |  openslr      | 4.3.0 |
 |  [PP-TTS-FastSpeech2](audio/speech_synthesis/fastspeech2/paddlepaddle)  |  PaddlePaddle  |  CSMSC        | 3.1.0 |
 |  [PP-TTS-HiFiGAN](audio/speech_synthesis/hifigan/paddlepaddle)          |  PaddlePaddle  |  CSMSC        | 3.1.0 |
 |  [Tacotron2](audio/speech_synthesis/tacotron2/pytorch)                  |  PyTorch       |  LJSpeech     | 2.2.0 |
diff --git a/audio/speech_synthesis/cosyvoice/deepspeed/README.md b/audio/speech_synthesis/cosyvoice/deepspeed/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..37c8e24754231d01a2f8e489c471743a6848a3a6
--- /dev/null
+++ b/audio/speech_synthesis/cosyvoice/deepspeed/README.md
@@ -0,0 +1,69 @@
+# CosyVoice2
+
+## Model Description
+
+CosyVoice2-0.5B is a small speech model designed to understand and generate human-like speech. It can be used for tasks like voice assistants, text-to-speech, or voice cloning. With 0.5 billion parameters, it is lightweight and works well on devices with limited computing power. It focuses on natural-sounding voices and easy customization.
+
+## Supported Environments
+
+| GPU    | [IXUCA SDK](https://gitee.com/deep-spark/deepspark#%E5%A4%A9%E6%95%B0%E6%99%BA%E7%AE%97%E8%BD%AF%E4%BB%B6%E6%A0%88-ixuca) | Release |
+| :----: | :----: | :----: |
+| BI-V150 | 4.3.0     |  25.09  |
+
+## Model Preparation
+
+### Prepare Resources
+
+Pretrained model: <https://huggingface.co/FunAudioLLM/CosyVoice2-0.5B>
+
+Dataset: <https://openslr.elda.org/resources/60/>
+
+```bash
+mkdir -p /root/datasets/openslr/libritts
+cd /root/datasets/openslr/libritts
+wget https://openslr.elda.org/resources/60/train-clean-100.tar.gz
+wget https://openslr.elda.org/resources/60/dev-clean.tar.gz
+tar -xvzf train-clean-100.tar.gz
+tar -xvzf dev-clean.tar.gz
+```
+
+### Install Dependencies
+
+Contact the Iluvatar administrator to get the missing packages:
+
+- transformers-4.45.2+corex.4.3.0-py3-none-any.whl
+- deepspeed-0.16.4+corex.4.3.0-cp310-cp310-linux_x86_64.whl
+
+```bash
+pip3 install -r requirements.txt
+git clone --recursive https://github.com/FunAudioLLM/CosyVoice.git
+# If you failed to clone the submodule due to network failures, please run the following command until success
+cd CosyVoice
+git submodule update --init --recursive
+
+mkdir -p pretrained_models
+# download CosyVoice2-0.5B model into pretrained_models dir
+
+# If you encounter sox compatibility issues
+# ubuntu
+sudo apt-get install sox libsox-dev
+# centos
+sudo yum install sox sox-devel
+```
+
+## Model Training
+
+```bash
+cp ../run_dpo.sh examples/libritts/cosyvoice2/
+cp ../run.sh examples/libritts/cosyvoice2/
+cd examples/libritts/cosyvoice2/
+
+export PYTHONPATH=../../../:../../../third_party/Matcha-TTS:$PYTHONPATH
+bash run.sh
+```
+
+## Model Results
+
+## References
+
+- [CosyVoice](https://github.com/FunAudioLLM/CosyVoice/commit/0a496c18f78ca993c63f6d880fcc60778bfc85c1)
\ No newline at end of file
diff --git a/audio/speech_synthesis/cosyvoice/deepspeed/requirements.txt b/audio/speech_synthesis/cosyvoice/deepspeed/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..f3f0917dac4757f45b359a0514f0cda2169770e9
--- /dev/null
+++ b/audio/speech_synthesis/cosyvoice/deepspeed/requirements.txt
@@ -0,0 +1,40 @@
+--extra-index-url https://download.pytorch.org/whl/cu121
+--extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/ # https://github.com/microsoft/onnxruntime/issues/21684
+conformer==0.3.2
+#deepspeed==0.15.1; sys_platform == 'linux'
+diffusers==0.29.0
+fastapi==0.115.6
+fastapi-cli==0.0.4
+gdown==5.1.0
+gradio==5.4.0
+grpcio==1.57.0
+grpcio-tools==1.57.0
+hydra-core==1.3.2
+HyperPyYAML==1.2.2
+inflect==7.3.1
+librosa==0.10.2
+lightning==2.2.4
+matplotlib==3.7.5
+modelscope==1.20.0
+networkx==3.1
+omegaconf==2.3.0
+#onnx==1.16.0
+#onnxruntime-gpu==1.18.0; sys_platform == 'linux'
+onnxruntime==1.18.0; sys_platform == 'darwin' or sys_platform == 'win32'
+openai-whisper
+protobuf==4.25
+pyarrow==18.1.0
+pydantic==2.7.0
+pyworld==0.3.4
+rich==13.7.1
+soundfile==0.12.1
+tensorboard==2.14.0
+#tensorrt-cu12==10.0.1; sys_platform == 'linux'
+#tensorrt-cu12-bindings==10.0.1; sys_platform == 'linux'
+#tensorrt-cu12-libs==10.0.1; sys_platform == 'linux'
+#torch==2.3.1
+#torchaudio==2.3.1
+#transformers==4.40.1
+uvicorn==0.30.0
+wetext==0.0.4
+wget==3.2
\ No newline at end of file
diff --git a/audio/speech_synthesis/cosyvoice/deepspeed/run.sh b/audio/speech_synthesis/cosyvoice/deepspeed/run.sh
new file mode 100644
index 0000000000000000000000000000000000000000..cc0b7c2e06372d7b7f654f1370875a91a09ec205
--- /dev/null
+++ b/audio/speech_synthesis/cosyvoice/deepspeed/run.sh
@@ -0,0 +1,126 @@
+#!/bin/bash
+# Copyright (c) 2025, Shanghai Iluvatar CoreX Semiconductor Co., Ltd.
+# All Rights Reserved.
+#
+# 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.
+
+# Copyright 2024 Alibaba Inc. All Rights Reserved.
+. ./path.sh || exit 1;
+
+stage=-1
+stop_stage=5
+
+data_url=www.openslr.org/resources/60
+data_dir=/root/datasets/openslr/libritts
+pretrained_model_dir=../../../pretrained_models/CosyVoice2-0.5B
+
+if [ ${stage} -le -1 ] && [ ${stop_stage} -ge -1 ]; then
+  echo "Data Download"
+  #for part in dev-clean train-clean-100; do
+  #  local/download_and_untar.sh ${data_dir} ${data_url} ${part}
+  #done
+fi
+
+if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
+  echo "Data preparation, prepare wav.scp/text/utt2spk/spk2utt"
+  for x in train-clean-100 dev-clean; do
+    mkdir -p data/$x
+    python3 local/prepare_data.py --src_dir $data_dir/LibriTTS/$x --des_dir data/$x
+  done
+fi
+
+if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
+  echo "Extract campplus speaker embedding, you will get spk2embedding.pt and utt2embedding.pt in data/$x dir"
+  for x in train-clean-100 dev-clean; do
+    python3 tools/extract_embedding.py --dir data/$x \
+      --onnx_path $pretrained_model_dir/campplus.onnx
+  done
+fi
+
+if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
+  echo "Extract discrete speech token, you will get utt2speech_token.pt in data/$x dir"
+  for x in train-clean-100 dev-clean; do
+    python3 tools/extract_speech_token.py --dir data/$x \
+      --onnx_path $pretrained_model_dir/speech_tokenizer_v2.onnx
+  done
+fi
+
+if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
+  echo "Prepare required parquet format data, you should have prepared wav.scp/text/utt2spk/spk2utt/utt2embedding.pt/spk2embedding.pt/utt2speech_token.pt"
+  for x in train-clean-100 dev-clean; do
+    mkdir -p data/$x/parquet
+    python3 tools/make_parquet_list.py --num_utts_per_parquet 1000 \
+      --num_processes 10 \
+      --src_dir data/$x \
+      --des_dir data/$x/parquet
+  done
+fi
+
+# train llm
+export CUDA_VISIBLE_DEVICES="0,1,2,3"
+num_gpus=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}')
+job_id=1986
+dist_backend="nccl"
+num_workers=2
+prefetch=100
+train_engine=torch_ddp
+if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
+  echo "Run train. We only support llm traning for now"
+  if [ $train_engine == 'deepspeed' ]; then
+    echo "Notice deepspeed has its own optimizer config. Modify conf/ds_stage2.json if necessary"
+  fi
+  cat data/train-clean-100/parquet/data.list > data/train.data.list
+  cat data/dev-clean/parquet/data.list > data/dev.data.list
+  # NOTE will update llm/hift training later
+  for model in llm; do
+    torchrun --nnodes=1 --nproc_per_node=$num_gpus \
+        --rdzv_id=$job_id --rdzv_backend="c10d" --rdzv_endpoint="localhost:1234" \
+      cosyvoice/bin/train.py \
+      --train_engine $train_engine \
+      --config conf/cosyvoice2.yaml \
+      --train_data data/train.data.list \
+      --cv_data data/dev.data.list \
+      --qwen_pretrain_path $pretrained_model_dir/CosyVoice-BlankEN \
+      --model $model \
+      --checkpoint $pretrained_model_dir/$model.pt \
+      --model_dir `pwd`/exp/cosyvoice2/$model/$train_engine \
+      --tensorboard_dir `pwd`/tensorboard/cosyvoice2/$model/$train_engine \
+      --ddp.dist_backend $dist_backend \
+      --num_workers ${num_workers} \
+      --prefetch ${prefetch} \
+      --pin_memory \
+      --use_amp \
+      --deepspeed_config ./conf/ds_stage2.json \
+      --deepspeed.save_states model+optimizer
+  done
+fi
+
+# average model
+average_num=5
+if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
+  for model in llm flow hifigan; do
+    decode_checkpoint=`pwd`/exp/cosyvoice/$model/$train_engine/${model}.pt
+    echo "do model average and final checkpoint is $decode_checkpoint"
+    python3 cosyvoice/bin/average_model.py \
+      --dst_model $decode_checkpoint \
+      --src_path `pwd`/exp/cosyvoice/$model/$train_engine  \
+      --num ${average_num} \
+      --val_best
+  done
+fi
+
+if [ ${stage} -le 7 ] && [ ${stop_stage} -ge 7 ]; then
+  echo "Export your model for inference speedup. Remember copy your llm or flow model to model_dir"
+  python3 cosyvoice/bin/export_jit.py --model_dir $pretrained_model_dir
+  python3 cosyvoice/bin/export_onnx.py --model_dir $pretrained_model_dir
+fi
\ No newline at end of file
diff --git a/tests/model_info.json b/tests/model_info.json
index 87f01851f7d60254bfa04ee050b006be947453a4..12bb3d93905b5a0b53886273cda704ba317f7022 100644
--- a/tests/model_info.json
+++ b/tests/model_info.json
@@ -7511,6 +7511,30 @@
             "github_branch": "",
             "github_path": "",
             "priority": "P4"
+        },
+        {
+            "model_name": "cosyvoice",
+            "framework": "pytorch",
+            "release_version": "25.09",
+            "release_sdk": "4.3.0",
+            "release_gpgpu": "BI-V150",
+            "latest_sdk": "",
+            "latest_gpgpu": "",
+            "category": "audio/speech_synthesis",
+            "toolbox": "",
+            "mdims": "",
+            "dataset": "",
+            "license": "",
+            "model_path": "deepsparkhub/audio/speech_synthesis/cosyvoice/pytorch",
+            "readme_file": "deepsparkhub/audio/speech_synthesis/cosyvoice/pytorch/README.md",
+            "bitbucket_repo": "",
+            "bitbucket_branch": "",
+            "bitbucket_path": "",
+            "develop_owner": "",
+            "github_repo": "",
+            "github_branch": "",
+            "github_path": "",
+            "priority": "P4"
         }
     ]
 }
\ No newline at end of file