Although I get good result from training I want to know if adaptation is much better for a single speaker ?
I have around 32 sentences. I have recorded each sentence 10 times and train them. However could you tell me if its better to adapt than train ?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
You did not provide enough information to enable answer on this question:
1) Amount of data used for training and adaptation in hours. Trainign is reasonable starting from 5 hours of data.
2) The baseline model you used for adaptation. It depends if model is far from speech you want to recognize or close to it.
3) The example of the audio recording you want to adapt to. It helps to determine the above.
4) The type of language you want to recognize (isolated commands, dictation, etc). For more diverse language you need more training data and adaptation could work better.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Although I get good result from training I want to know if adaptation is much better for a single speaker ?
I have around 32 sentences. I have recorded each sentence 10 times and train them. However could you tell me if its better to adapt than train ?
You did not provide enough information to enable answer on this question:
1) Amount of data used for training and adaptation in hours. Trainign is reasonable starting from 5 hours of data.
2) The baseline model you used for adaptation. It depends if model is far from speech you want to recognize or close to it.
3) The example of the audio recording you want to adapt to. It helps to determine the above.
4) The type of language you want to recognize (isolated commands, dictation, etc). For more diverse language you need more training data and adaptation could work better.