fpga-open-speech-tools / simulink_models

Simulink models for speech and audio processing
MIT License
4 stars 4 forks source link

Open MHA: O1 AR Filter Optimization Fixed Point Output size increasing #56

Closed dack-fe closed 3 years ago

dack-fe commented 3 years ago

Because of the fixed point multiplication between the input and coefficients, the outputting signal is increasing. Do we want to convert this result to match the input size by removing the additional fractional bits?

o1_ar_filter_optimization.slx

matthewblunt commented 3 years ago

Yes, I think removing the fractional bits is the correct move. Because the signal is fed into the gain table afterword, the lost precision will not matter as the gain acquired from the table will be almost exactly the same.

dack-fe commented 3 years ago

Currently being handled with parameters set in init scripts