mahaotian

模型替换成Tacotron2时，推理出现问题

2

当我输入一句带有逗号的句子，例如“欢迎使用工具箱，现已支持中文输入”时，我将句子处理成两个子句（与博主的处理方法一致），那么输入的batch就是2。但在Tacotron2里面，在判断gate门输出的时候，如果输入的句子batch为1时，就可以判断是否调出解码；但输入的句子batch为2时，就不能判断成功，我能将gate的阈值也变成batch为2的矩阵吗？

FileNotFoundError: [Errno 2] Unable to synchronously open file

3

 I have move the h5 files to other place because of lack of memory, what can i do to read h5 files when cut.load_features() ?

[BUG]: the lengths of the features after FACodecEncoderV2 is not match

1

## bug of FACodecEncoderV2 I have extracted prosody_feature and encoder_output from FACodecEncoderV2. It raise wrong when I use fa_decoder_v2 to extract vq codecs becaucse the lengths of prosody_feature(torch.Size([1, 20, 281]))...

The model does not converge when I use hifi-codec to train NAR of valle. The data i used is a chinese dataset while its duration is 5000 hours. How can...

it takes too long for DynamicBucketingSampler to load state dict

5

When I retrained 30,000 hours of data from checkpoint, it took a long time to load state dict for DynamicBucketingSampler(more than 2 hours).It's it normal ? here is my code:...

OSError: [Errno 9] Unable to synchronously open file (unable to lock file, errno = 9, error message = 'Bad file descriptor')

1

I have use ddp to train a model with diffence machine. When I `load_features() ` while get datas from h5 files, I meet this error. How to fix this？

Why I only get 5 classes of emotion

3

According to the official description, the motion2vec_plus model should be able to categorize speech into 9 classes, but why did I actually run it with only 5 classes（angry、happy、neutral、sad and ）?...