-
Hello,
I am interested in getting the word level confidences from Quartznet model. What would be the best approach to obtain some proxy values to word level confidences? Taking the logit sequences…
-
when i try to upload a pretrained model
quartznet = nemo_asr.models.EncDecCTCModel.from_pretrained(model_name="QuartzNet15x5Base-En")
it is giving error
AttributeError: module 'nemo.collections.…
-
**Is your feature request related to a problem? Please describe.**
Not a problem, just seeking some guidance on how to replicate the timestamp extraction code featured within the [Offline_ASR](http…
-
The CER value during the validation of the training process does not align with the inference process. And during the inference process, the execution is like the following part.
`wer_value = word_er…
-
**Describe the bug**
Hi. I cannot seem to get the right syntax for the quartz_jobs.xml file. I always get this exception:
```
System.InvalidOperationException: was not expected.
```
I also see…
-
I'm using the code provided [here](https://github.com/NVIDIA/NeMo/blob/73875d874b006b672a8e407687ef414ef7a7b282/tutorials/asr/Offline_ASR.ipynb) to extract timestamps along with the transcription. The…
-
When will the RavenDB job store implementation be ready?
-
At the Minute the Service Activator is an IHostedService CommandHandlers processing these messages cannot consume a Scoped Service without Injecting a ScopeFactory and creating a new scope inside my H…
-
Hi Dominik,
these are some nice tests :+1: . I was wondering if you have any documentation on how you got Scribosermo running? It looks like performance and size are very good and I'd love to add i…
-
Hi,
Can I use Persian Multiple speaker audio datasets for fine-tuning QuartzNet15*5? Will be the WER low as I fine-tune it with a single speaker dataset?
Thanks
**Environment overview**
- En…