Instructions to use speechbrain/m-ctc-t-large with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use speechbrain/m-ctc-t-large with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="speechbrain/m-ctc-t-large")# Load model directly from transformers import AutoModelForCTC model = AutoModelForCTC.from_pretrained("speechbrain/m-ctc-t-large", dtype="auto") - Notebooks
- Google Colab
- Kaggle
Examples for recognition of Surzhik (a mix of Ukrainian and Russian languages)
I would like to share a link to my experiments about testing this model over Surzhik speech (a mix of Ukrainian and Russian languages).
It's an important topic for research in speech recognition of mixed languages so I hope this discussion will help people to find my experiments better.
Link to first examples: https://t.me/speech_recognition_uk/15720
Hi!
I would suggest to follow these steps:
1- Open a PR to add your recipe in SpeechBrain's github (https://github.com/speechbrain/speechbrain)
2- Once the code is fine, you can share with us the output folder of the best experiments. We will upload it in the google drive and to the official HF repository of SpeechBrain.
Does it sounds like a good plan for you?
No, no. Exactly m-ctc-t-large model did recognition so there's no need to create an independent model for Surzhik.