-
请问在aishell3-test上的CER的计算设置是什么呢? 我用CosyVoice-300M的inference_cross_lingual 推理计算了aishell3-test的CER,但是算的结果是8.1%,和论文的3.8%还有一定距离?是不是我的设置有问题?
-
大佬们,第一次使用ncnn,在转换模型的时候报错
阅读源码发现部分环节会直接随机初始化输入到模型,但是模型的输入是有范围的,请问这个有什么方法可以解决吗?
## error log | 日志或报错信息 | ログ
![image](https://github.com/user-attachments/assets/353e210c-b45b-4f62-b5ad-fa51db1a47df)…
-
目前测试下来mb_melgan_csmsc的推理速度满足离线环境下CPU实时推理,其他的VOC模型都太慢,但是我用fastspeech2_mix微调克隆了一个男生音色后,使用mb_melgan_csmsc推理音色就不对,只能使用aishell3数据集训练的VOC模型,而mb_melgan只有csmsc数据集训练的模型,问题来了,如何使用aishell3数据训练一个mb_melgan模型,求指点
-
![image](https://user-images.githubusercontent.com/52569141/183853296-9f0dfad0-dbdc-44de-9a20-e9ee9e4e03d4.png)
after training 200k steps, train loss and valid loss increased rapidly. anyone encounte…
-
1.i use [7b_tiva_v0](https://huggingface.co/ChocoWu/nextgpt_7b_tiva_v0)
2.chage NExT-GPT-Lagacy/code/config/stage_3.yaml:
```
freeze_lm: true
freeze_input_proj: true
freeze_output_proj: true
```…
-
(CDFSE) root@autodl-container-66ee44be9a-c385597c:~/CDFSE_FastSpeech2-main# python3 preprocess.py config/AISHELL3/preprocess.yaml
Processing Data ...
0%| …
-
I plan to fine-tune my own dataset based on the AISHELL3 model, but my dataset only has 6 speakers, while AISHELL3 has 218. When loading the model, an error occurred due to the size mismatch. Addition…
-
Hi, thanks for your great work. I train styletts2 on aishell3, after 50 epochs of stage2, I try to test the model, but got a weird pronunciation.
[example.zip](https://github.com/user-attachments/fi…
-
尝试使用aishell3的onnx转换脚本,报错如下:
warnings.warn("torch.nn.utils.weight_norm is deprecated in favor of torch.nn.utils.parametrizations.weight_norm.")
Traceback (most recent call last):
File "/media/nvm…
-
./run.sh --stage 0 --stop_stage 0
执行至 /workspace/paddlespeech/paddlespeech/t2s/exps/fastspeech2/preprocess.py 287行,提示aishell3 路径缺少/aishell3/train/wav路径,aishell3数据集是从官网下载的,需要怎么样进行划分呢? 这个preprocess.py…