onnxruntime加速 #44
Description
[load_quantize_encoder] start. model_path='ckpt/speech_tokenizer'
Configured for 24kHz frontend.
2025-12-25 18:04:58.332369081 [E:onnxruntime:Default, provider_bridge_ort.cc:1992 TryGetProviderInfo_CUDA] /onnxruntime_src/onnxruntime/core/session/provider_bridge_ort.cc:1637 onnxruntime::Provider& onnxruntime::ProviderLibrary::Get() [ONNXRuntimeError] : 1 : FAIL : Failed to load library libonnxruntime_providers_cuda.so with error: libcudnn.so.9: cannot open shared object file: No such file or directory
2025-12-25 18:04:58.332398317 [W:onnxruntime:Default, onnxruntime_pybind_state.cc:965 CreateExecutionProviderInstance] Failed to create CUDAExecutionProvider. Require cuDNN 9.* and CUDA 12.*. Please install all dependencies as mentioned in the GPU requirements page (https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements), make sure they're in the PATH, and that your GPU is supported.
2025-12-25 18:04:58,514 WETEXT INFO building fst for zh_normalizer ...
INFO: building fst for zh_normalizer ...
2025-12-25 18:05:18,547 WETEXT INFO done
INFO: done
2025-12-25 18:05:18,547 WETEXT INFO fst path: /data/shengzewen/conda_env/szw_glmtts/lib/python3.10/site-packages/tn/zh_tn_tagger.fst
INFO: fst path: /data/shengzewen/conda_env/szw_glmtts/lib/python3.10/site-packages/tn/zh_tn_tagger.fst
2025-12-25 18:05:18,547 WETEXT INFO /data/shengzewen/conda_env/szw_glmtts/lib/python3.10/site-packages/tn/zh_tn_verbalizer.fst
INFO: /data/shengzewen/conda_env/szw_glmtts/lib/python3.10/site-packages/tn/zh_tn_verbalizer.fst
2025-12-25 18:05:18,552 WETEXT INFO found existing fst: /data/shengzewen/conda_env/szw_glmtts/lib/python3.10/site-packages/tn/en_tn_tagger.fst
INFO: found existing fst: /data/shengzewen/conda_env/szw_glmtts/lib/python3.10/site-packages/tn/en_tn_tagger.fst
2025-12-25 18:05:18,552 WETEXT INFO /data/shengzewen/conda_env/szw_glmtts/lib/python3.10/site-packages/tn/en_tn_verbalizer.fst
INFO: /data/shengzewen/conda_env/szw_glmtts/lib/python3.10/site-packages/tn/en_tn_verbalizer.fst
2025-12-25 18:05:18,552 WETEXT INFO skip building fst for en_normalizer ...
INFO: skip building fst for en_normalizer ...
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 89.47it/s]
Loading HiFT model from ckpt/hift/hift.pt on cuda...
感谢开源这么好的项目！
我推理成功了，但是日志显示没有用到onnxruntime加速GPU，请问这个该如何解决。
我试过onnxruntime==1.19.0和1.19.2都是这个报错，当前环境是python 3.10，cuda 12.8
速度大概是1:1(生成的音频时长：推理时长)，这个是正常的吗