Given a audio that speaker mainly speaks 90% of time in Vietnamese, 10% of time in English. I've tested your model with this type of audio and English words are interpreted as Vietnamese language.
I am thinking of re-train the model with a dataset that contains the English and Vietnamese in transcript. Do you think this approach is feasible or not ?
Hi VinAI Team,
Given a audio that speaker mainly speaks 90% of time in Vietnamese, 10% of time in English. I've tested your model with this type of audio and English words are interpreted as Vietnamese language.
I am thinking of re-train the model with a dataset that contains the English and Vietnamese in transcript. Do you think this approach is feasible or not ?
Thank you