-
I am getting nan outputs from the encoder of pruned transducer streaming model.
tensor([[[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan, ..., nan, nan, nan],
[nan, nan, nan…
-
```
Hi, since I could not find a mailinglist, I am going to post this here.
When I try to convert a mdb that even mdbtools dosn't oepn (without error
anyways), I get an:
Exception in thread "main" …
-
```
Hi, since I could not find a mailinglist, I am going to post this here.
When I try to convert a mdb that even mdbtools dosn't oepn (without error
anyways), I get an:
Exception in thread "main" …
-
Things like ordinal numbers, dates, and times are not being pronounced properly. They come out as a quick 'fft' noise. This is happening for both intents and on the test page using the speak button.
…
-
## 🐛 Bug
I made my own wav2vec2.0 model, and pretrained it with some data.
model training was successful, and inference is working with the command written in wav2vec2.0 README.md.
But the problem …
-
```
Hi, since I could not find a mailinglist, I am going to post this here.
When I try to convert a mdb that even mdbtools dosn't oepn (without error
anyways), I get an:
Exception in thread "main" …
-
```
Hi, since I could not find a mailinglist, I am going to post this here.
When I try to convert a mdb that even mdbtools dosn't oepn (without error
anyways), I get an:
Exception in thread "main" …
-
```
Hi, since I could not find a mailinglist, I am going to post this here.
When I try to convert a mdb that even mdbtools dosn't oepn (without error
anyways), I get an:
Exception in thread "main" …
-
**Describe the bug**
There is a mismatch in the number of speech references and the number of speakers (which is 2 for the Librimix dataset).
Because of this issue, we cannot run the recipe training…
-
## 🐛 Bug
I am trying to obtain inference for sample flac files (in a directory) using the 'infer.py' script in fairseq.
### To Reproduce
I installed fairseq using the following commands:
```…