-
**Describe the bug**
在构建cosyvoice类时候出现问题,对应这句cosyvoice = CosyVoice('pretrained_models/CosyVoice-300M-SFT')
报错内容,>>> cosyvoice = CosyVoice('pretrained_models/CosyVoice-300M-SFT')
/home/ialover/min…
-
## 🐛 Bug
### To Reproduce
Steps to reproduce the behavior (**always include the command you ran**):
麻烦大佬有空时能解答解答,十分感谢。
基于/runtime/python/http/server.py的代码做了简单的修改,具体的代码会在下面贴出。
存在问题:识别多个离线音…
-
在我的音频中老师的提问和学生的回答时间间隔为1秒左右, 我减少max_end_silence_time为500ms尝试精确定位句尾,但是没有效果,无法精确分离老师和学生的话,请问还可以尝试什么配置呢?
**vad_model模型配置如下:**
```
frontend: WavFrontendOnline
frontend_conf:
fs: 16000
window:…
-
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)
## 🐛 Bug
### To Reproduce
Steps to reproduce the behavio…
-
## 🐛 Bug
punc入参的文本为空时报错如下:
` Expected tensor for argument #1 'indices' to have one of the following scalar types: Long, Int; but got torch.cuda.DoubleTensor instead (while checking arguments for…
-
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)
## 🐛 Bug
使用1.x版本funasr,跑aishell训练例子时,在stage 1 compute_audio_cmv…
-
ubuntu
git clone https://github.com/alibaba-damo-academy/FunASR
```python
from funasr import AutoModel
# paraformer-zh is a multi-functional asr model
# use vad, punc, spk or not as you need
m…
-
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)
## 🐛 Bug
我用一个语音离线生成spk,但是发现key与sentence_info内容不能匹配,sentence_inf…
-
调整了一下speakerlab/bin/infer_sv.py的推理代码,使用的模型是[https://www.modelscope.cn/models/iic/speech_eres2net_large_sv_en_voxceleb_16k/files](url),测试时听起来非常像的两个音频,其说话人相似度偏低。以0_s.wav和0_t.wav为例,取前3s的音频测试的相似度最高,为0.177…
-
```
from modelscope.pipelines import pipeline
sv_pipeline = pipeline(
task='speaker-verification',
model='damo/speech_campplus_sv_zh-cn_16k-common',
model_revision='v1.0.0'
)
speake…