Parcollet Titouan

@anautsch this looks like something is not working as intended on the profiling tests ?

Hi, this should not cause any issue. Are you on the latest SpeechBrain develop ?

This error will disappear once we release a new SB version. But we will merge it if the tests pass so we don't have any issues anymore

@kenro515 Hi ! Any chance that you can solve the pre-commit test failing ? (Just a formatting issue). Then I will merge ! Thanks !

Hi, I encourage you to first read properly the tutorial we provide on our website for inference. It is most likely that you did not create the pretrainer in your...

Hi, did you try to follow this tutorial: https://colab.research.google.com/drive/1UwisnAjr8nQF3UnrkIJ4abBMAWzVwBMh?usp=sharing ?

Hi, this is a question for @mravanelli and @anautsch

Hi @helloworld02202, sorry for the late reply, I only found time to look into the code again. This function must return the mask i.e. will need to retrieve the max...

Hi, Indeed, models are usually trained with fixed bit rates that are connected to the training dataset. I am not sure about VoxCeleb but you will certainly find this information...

Hi and thanks ! I see that you decided to go for BPE and seq2seq. For some reasons, w2v2-based ASR models tend to work better with pure CTC and characters...