Update Calibration parameters

Added Calibration Low and High Cutoff Freq. (Hz) options to advanced settings to control the frequency range of diffuse field calibration

Author: ShanonPearce

ash_toolset/__main__.py

@@ -2634,6 +2634,22 @@ def get_screen_size(default_width=1722, default_height=717):
                                     with dpg.tooltip("octave_smoothing_n"):
                                         dpg.add_text("Sets the fractional octave smoothing for calibration filters (e.g., 3 = 1/3 octave). Higher values result in more detail; lower values result in a smoother curve.")
 
+                                    
+                                # Low Cutoff Frequency
+                                with dpg.table_row():
+                                    dpg.add_text("Binaural Sim.")
+                                    dpg.add_text("Calibration Low Cutoff Freq. (Hz)")
+                                    dpg.add_input_int(label="", width=190, tag='brir_df_cal_low_f', min_value=1, max_value=500,default_value=loaded_values["brir_df_cal_low_f"],min_clamped=True, max_clamped=True)
+                                    with dpg.tooltip("brir_df_cal_low_f"):
+                                        dpg.add_text("The lower frequency limit for inversion. Below this point, the correction magnitude is clamped to the value at this limit to avoid over-excursion.")
+                                
+                                # High Cutoff Frequency
+                                with dpg.table_row():
+                                    dpg.add_text("Binaural Sim.")
+                                    dpg.add_text("Calibration High Cutoff Freq. (Hz)")
+                                    dpg.add_input_int(label="", width=190, tag='brir_df_cal_high_f', min_value=5000, max_value=22050,default_value=loaded_values["brir_df_cal_high_f"],min_clamped=True, max_clamped=True)
+                                    with dpg.tooltip("brir_df_cal_high_f"):
+                                        dpg.add_text("The upper frequency limit for inversion. Above this point, the magnitude is extended (clamped) to prevent sharp high-frequency rises or artifacts.")    
 
                                 with dpg.table_row():
                                     dpg.add_text("Binaural Sim.")

ash_toolset/brir_generation.py

@@ -85,6 +85,8 @@ def generate_integrated_brir(brir_name,  spatial_res=1, report_progress=0, gui_l
             octave_smoothing_n=brir_meta_dict.get("octave_smoothing_n")
             brir_df_cal_factor=brir_meta_dict.get("brir_df_cal_factor")
             gen_fir_length=brir_meta_dict.get("gen_fir_length")
+            brir_df_cal_low_f=brir_meta_dict.get("brir_df_cal_low_f")
+            brir_df_cal_high_f=brir_meta_dict.get("brir_df_cal_high_f")
         else:
             raise ValueError('brir_meta_dict not populated')
 
@@ -725,7 +727,7 @@ def generate_integrated_brir(brir_name,  spatial_res=1, report_progress=0, gui_l
         #convert to mag
         brir_fft_avg_mag = hf.db2mag(brir_fft_avg_db)      
         #level ends of spectrum
-        brir_fft_avg_mag_sm = hf.level_spectrum_ends(brir_fft_avg_mag, 15, 18500, smooth_win = octave_smoothing_n, n_fft=CN.N_FFT)#40, 19000, smooth_win = 7 
+        brir_fft_avg_mag_sm = hf.level_spectrum_ends(brir_fft_avg_mag, brir_df_cal_low_f, brir_df_cal_high_f, smooth_win = octave_smoothing_n, n_fft=CN.N_FFT)#40, 19000, smooth_win = 7 
         #octave smoothing
         brir_fft_avg_mag_sm = hf.smooth_gaussian_octave(data=brir_fft_avg_mag_sm, n_fft=CN.N_FFT, fraction=octave_smoothing_n)
 
@@ -757,7 +759,7 @@ def generate_integrated_brir(brir_name,  spatial_res=1, report_progress=0, gui_l
             # 2. Run your function with the override
             # This will return the EQ-IR as a (1, 1, n_fft) array
             eq_ir_dataset = hf.equalize_brirs_parametric(brir_dataset=impulse, diff_db_override=hf.mag2db(brir_fft_avg_mag_inv), override_n_fft=CN.N_FFT, 
-                                                         num_filters=50,low_freq_cut = 12.0, high_freq_cut = 18000.0)
+                                                         num_filters=50,low_freq_cut = brir_df_cal_low_f, high_freq_cut = brir_df_cal_high_f)
             # 3. Extract the 1D array for easy convolution later
             brir_df_inv_fir = eq_ir_dataset.flatten()
         else:

ash_toolset/callbacks.py

@@ -3172,6 +3172,8 @@ def get_brir_dict():
     hp_rolloff_comp = dpg.get_value('hp_rolloff_comp')
     fb_filtering = dpg.get_value('fb_filtering')
     brir_max_length = dpg.get_value('brir_max_length')
+    brir_df_cal_low_f = dpg.get_value('brir_df_cal_low_f')
+    brir_df_cal_high_f = dpg.get_value('brir_df_cal_high_f')
 
 
     brir_meta_dict = {
@@ -3226,7 +3228,7 @@ def get_brir_dict():
         'room_target': room_target, 'room_target_int': room_target_int, 'direct_gain_db': direct_gain_db, 
         'ac_space_short': ac_space_short,   'ac_space_gui': ac_space_gui, 'brir_hp_comp': brir_hp_comp, 'brir_elf':brir_elf,
         'samp_freq_int': samp_freq_int, 'samp_freq_str': samp_freq_str,   
-        'bit_depth': bit_depth, 'bit_depth_str': bit_depth_str, 
+        'bit_depth': bit_depth, 'bit_depth_str': bit_depth_str, 'brir_df_cal_low_f': brir_df_cal_low_f, 'brir_df_cal_high_f': brir_df_cal_high_f, 
 
         # Additional variables
         'hrtf_symmetry': hrtf_symmetry, 'er_delay_time': er_delay_time, 'reverb_tail_crop_db': reverb_tail_crop_db, 'brir_df_cal_mode':brir_df_cal_mode, 'brir_max_length':brir_max_length,

ash_toolset/constants.py

@@ -1589,6 +1589,8 @@ def to_circular(angle):
     "octave_smoothing_n": 6,
     "brir_df_cal_factor":1.0,
     "gen_fir_length":4096,
+    "brir_df_cal_low_f":10,
+    "brir_df_cal_high_f":19500,
 
     # AS Import tool
     "as_reverb_tail_mode": AS_TAIL_MODE_LIST[0],

ash_toolset/helper_functions.py

@@ -1963,6 +1963,8 @@ def equalize_brirs_parametric(
     and applies filtering via FFT convolution for massive speed gains.
     """
     # --- 1. Preparation ---
+    low_freq_cut = float(low_freq_cut)
+    high_freq_cut = float(high_freq_cut)
     brir_dataset = np.asarray(brir_dataset, dtype=np.float64)
     orig_shape = brir_dataset.shape
     n_samples = orig_shape[-1]