-
我们想使用 FunASR 做呼叫中心的 录音质检
发现同一段录音,使用:
funasr ++model=paraformer-zh ++vad_model="fsmn-vad" ++punc_model="ct-punc" ++input=测试录音文件.wav
---》
正确率还不错,基本都正确
但是,使用:中文离线文件转写服务(CPU版本)版本的
python3 funa…
-
Traceback (most recent call last):
File "E:\剪辑筛选\funclip\launch.py", line 6, in
funasr_model = AutoModel(model="iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch",
…
-
```
from funasr import AutoModel
# paraformer-zh is a multi-functional asr model
# use vad, punc, spk or not as you need
model = AutoModel(model="paraformer-zh", vad_model="fsmn-vad", punc_model…
-
版本信息:
funasr 0.7.4
torch 1.11.0+cu113
modelscope 1.7.1
错误信息:
CRITICAL: 08-17 03:17:59: paraformer_large_long_audio.py:152 * 139792295642880 Trace…
-
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
### Before asking:
1. search the issue…
-
server:
E:\FunASR\Test>python funasr_wss_server.py
model loading
2024-05-23 15:51:03,247 - modelscope - INFO - PyTorch version 2.3.0 Found.
2024-05-23 15:51:03,249 - modelscope - INFO - Loading as…
-
## 🐛 Bug
识别出来subword
茂名口音,
gt : 好 啲 呢 我 觉 得
pred: ho@@ al@@ ding ne@@ un@@ qu@@ ar@@ ter a
[2-28-2_00751262_00752898.zip](https://github.com/alibaba-damo-academy/FunASR/files/1452228…
-
做项目遇到并发报错,跑到FunAsr Demo里也去执行了下,发现了同样的错误,如下:
OS: [e.g. linux] ubuntu20.04 和 Macbook都测试了
Python Version:py37-torch1.13.1
Package Version:modelscope 1.4.1 funasr 0.3.1
Model:damo/speech_paraforme…
-
windows11
pytorch:2.2.1+cu118
python:3.8
使用pip安装funasr:1.0.19版本
使用pip安装modelscope:1.13.2版本
然后按着FunASR社区软件包Windows版本流程git下载以下文件:
```
mkdir -p D:/funasr-runtime-resources/models && cd D:/funasr-…
-
我想在`iic/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch`模型基础上进行训练,但是我的训练的音频数据是8000采样率,该如何修改