how to train the "little" model

[battlefor]

hello,I glad to obtain the amazing rnnoise lib,it is very suitable for me.

now i want to training the "little" model ,just like the "little" model provided with half as an alternative

i had study the rnnoise_data_little.c for a long time, but i still don't know how to get the rnnoise_data_little.c from my train.

[BancoLin]

@battlefor The rnnoise training script already covers 'little model' training steps (model sparsification), but your training data must large enough to trigger it. If I recall correctly, the little model training steps start at iteration 2500 and stop at iteration 8000. You can check your training log to see if the training runs beyond 2500 iterations.

If you want to fine tune the sparsification parameters, please go to check rnnoise/rnnoise.py. There are sparsify_start, sparsify_stop, sparsify_interval, and sparsify_exponent. You can control sparsify_start parameter to control when to start model sparsification.