-
From @vesis84 in #4543
> i just noticed there are 2 very similar files:
> steps/nnet3/align_lats.sh and steps/nnet3/chain/align_lats.sh
Check which of them, and how they are used (if used), and d…
-
After generate recording.jsonl.gz and supervison.jsonl.gz from kaldi data directory with command:
`lhost kaldi import `
I creat a dataset from these two manifests with:
```
recordings = load_man…
-
I fail to 'make' for kaldi/src. There is a bug in kaldi/src/cudamatrix/cublas-wrappers.h related to the last update (support for Cuda 11). I am using Cuda 10.2 and the error says CUSPARSE_SPMM_C…
-
Kaldi version: 5-5.636
CUDA support: yes
Driver Version: 440.95.01
CUDA Version: 10.2
OS: CentOS 7 x64
Virtualization: openVZ
NVIDIA Hardware: 2 x NVIDIA Tesla T4
All strated from unpredictab…
-
Hi everyone,
First of all, I am totally new in the ASR scene and english is not my mother language so I apologize.
So I understood that if a model has a Gr.fst it's a dynamic one and that's mean t…
-
When I ran the GigaSpeech recipe, I encountered this error:
```
======Prepare Dictionary START | current time : 2021-10-25-07:18:52====
local/prepare_dict.sh: Downloading and preparing CMUdict
…
-
cuda: 11.3, pytorch: 1.10.1, python: 3.8.5
I trained a conformer transducer model using aishell recipe, but got this as follows, could anyone give me a hand?
2022-01-08 19:48:09,317 (asr_train:563)…
-
Hi,
As I asked last time I am using the egs/ksponspeech/asr example. As you pointed out, I'm using the quantization option implemented in espnet1's asr_recog.py. However, what I applied does not see…
-
Hello @danpovey
It is obeserved from chain model output that the "output probability" is kind of flat, compared to CTC sharp peak .
I wonder how include-log-softmax=true would affect model perfor…
-
**Bug description**
I want to train tacotron2 on new data and at first I just try to run [2_TTS_Tacotron2_Training.ipynb](https://github.com/NVIDIA/NeMo/blob/r1.0.0rc1/tutorials/tts/2_TTS_Tacotron2…