-
Hey @dogancan
Thanks for updating the offset value in the decoding method using endpointing in `nnet3-online-recognizer.py`.
It works a lot better but I still run into trouble with getting the b…
-
When installing the kaldi as ```kaldi/tools/INSTALL```, I typed ```sudo make``` and met the following problem:
```
extras/check_dependencies.sh
extras/check_dependencies.sh: all OK.
tar xojf s…
-
Καλησπέρα, θα ήθελα να ρωτήσω αν ο οδηγός (του Kaldi) που έχετε για εγκατάσταση του gcc 6 σε Ubuntu 18.04 ισχύει και για Ubuntu 20.04 ή αν πρέπει να ακολουθήσουμε διαφορετική διαδικασία. Επίσης φαίνετ…
-
Hello, I am trying to run compile-train-graphs-without-lexicon and am getting an error saying that there is no lexical item for a label that matches my first wav file name. This makes me think that I …
-
WSL2 = Windows Subsystem for Linux,
Now WSL2 is basically the same as ubuntu and can also run docker.
I loaded the image of https://github.com/sskorol/vosk-api-gpu on WSL2, but it said that the GPU …
-
When 120 seconds of wav data was input to the voice enhancement task, the following error occurred in Stage 4.
The error did not occur with 1 second of wav data.
> 2022-03-14T16:05:07 (enh.sh:347:ma…
-
**Describe the bug**
Hello, when i use the deafult recipe for mustc v1, a bug raised(at stage 1):
**utils/validate_data_dir.sh: text contains 1 lines with non-printable characters**
I think this…
-
It has been a very long issue of Kaldi that, Kaldi is using tag "kaldi" to build Phonetisaurus-g2p. However, in modern OS (such as Ubuntu 18 or 20), original default of "python==python2" has alrea…
-
https://twitter.com/alphacep/status/1478804751382990853
-
I have a trained acoustic model (Pytorch RNN) that outputs frame wise phone likelihoods, and a fst composed of a token, lexicon, and grammar fst. I have tried to use pykaldi to connect my RNN and FST …