-
刚才下载了想在Windows On ARM平台上编译,遇到CMake配置时报错:
```
-- CMAKE_SYSTEM_NAME: Windows
-- CMAKE_SYSTEM_PROCESSOR: ARM64
-- location_onnxruntime_header_dir: location_onnxruntime_header_dir-NOTFOUND
-- locat…
-
关键词识别模型加载时间需要30秒!如何优化 ? Keyword spotter model loading time takes 30 seconds! How to optimize?
`/home/suser/sherpa-onnx-1.10.20/sherpa-onnx/c-api/c-api.cc:SherpaOnnxCreateKeywordSpotter:637 Keyword…
-
您好,我想训练四种语种,所以我将CAM++语种识别-中英粤日韩识别-16k当中的embedding model来初始化我的model,然后四种语种的训练时数分别为250h, 250h , 500h, 200h,每跑一个epoch 我就测试一次eval和test,test前一两个epoch acc有8成,但后面就逐渐往下掉6~7成,eval一直往上升到9成多,所以我认为是overfitting 。
…
-
在librispeech上测试model = AutoModel(model="paraformer-zh-streaming"),效果不理想
-
Notice: In order to resolve issues more efficiently, please raise issue following the template.
(注意:为了更加高效率解决您遇到的问题,请按照模板提问,补充细节)
## ❓ Questions and Help
理论上
SeACoParaformer-zh 支持热词
paraf…
-
运行环境:
操作系统:linux
python:3.8.16
modelscope:1.9.4
funasr: 0.8.4
gpu:T4 cuda:11.6
代码
```
from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks
from modelscope.…
-
我在 https://github.com/alibaba-damo-academy/FunASR/issues/908#issuecomment-1756610416 中看到回复说更新到 update modelscope-1.9.2后修复这个问题,
但是我在modelscope 1.9.4还是遇到了这个问题
asr = pipeline(
task=Tasks.auto_sp…
-
Log:
STDOUT: /home/runner/work/sherpa-onnx/sherpa-onnx/sherpa-onnx/csrc/offline-recognizer-sense-voice-impl.h:DecodeOneStream:260 Unknown language: en. Use 0 instead.
https://github.com/k2-fsa/she…
-
####question
按照https://github.com/alibaba-damo-academy/FunASR/tree/main/runtime/python/libtorch导出模型出错:
python3.8 -m funasr.export.export_model --model-name damo/speech_paraformer-large_asr_nat…
-
### 使用镜像:modelscope:ubuntu22.04-cuda11.8.0-py310-torch2.1.0-tf2.14.0-1.10.0
### 使用模型:OFA-MMSpeech语音识别-中文-aishell1-large
问题描述:
1、在模型“ofa_mmspeech_asr_aishell1_large_zh”上使用自己的wav上进行中文识别,识别的结果差别很大
…