剩下分割子分片处理过程

EmiyaEngine.py

@@ -9,10 +9,12 @@
 # 只要蘊藏著想成為真物的意志，偽物就比真物還要來得真實
 
 
+import math
 import numpy as np
 import scipy
 import librosa
 import resampy
+import random
 import multiprocessing
 
 
@@ -27,8 +29,9 @@ class EmiyaEngine:
     mid_signal_larray = ''
     mid_signal_rarray = ''
     mid_signal_carray = ''
-    mid_process_ponit = 32
-    mid_src_point = 2048
+    # 各进程处理参数
+    proc_mid_sr_rate = 32
+    proc_fft_length = 2048
     # 输出用参数
     output_file_path = ''
     output_sr = 0
@@ -91,12 +94,85 @@ def process_leader(self):
         self.save_file()
 
     def process_core(self, signal_piece, index):
+        # SRC 预处理
+        proc_mid_data = resampy.resample(signal_piece, 
+                                         self.input_sr, 
+                                         self.input_sr * self.proc_mid_sr_rate, 
+                                         filter='kaiser_fast')
+        # 加抖动处理
 
-
-        return [signal_piece, index]
+        # SRC 后处理
+        proc_fin_piece = resampy.resample(proc_mid_data, 
+                                          self.input_sr * self.proc_mid_sr_rate, 
+                                          self.output_sr, 
+                                          filter='kaiser_fast')
+        return [proc_fin_piece, index]
+
+    def find_threshold_point(self, input_fft):
+        # 鉴定频谱基本参数
+        amp_fft = abs(input_fft[range(self.proc_fft_length // 2)])
+        # Step0. 找出基波幅度
+        base_amp_freq = amp_fft.max()
+        # Step1. 找出接续的阈值
+        threshold_hit = 1.0e-11                                                      # 方差判定阈值
+        fin_threshold_point = 0                                                      # 最后的阈值点
+        find_range = int((self.proc_fft_length / 2) - 1)                             # 搜索的范围
+        start_find_pos = round(2000 / (self.input_sr / (self.proc_fft_length / 2)))  # 从2K频点附近开始寻找，加快速度
+        start_flag = True                                                            # 循环用的启动Flag
+        loop_count = 0                                                               # 循环计数器
+        legal_freq = (self.input_sr / 2) - 500                                       # 判定结果合法的阈值频率
+        forward_freq = 3000                                                          # 前向修正频率
+        order_freq = (self.input_sr / 2) - 6000                                      # 钦定频率
+        # Rev.1: 检查接续点是否符合常理
+        while start_flag or fin_threshold_point > round(legal_freq / (self.input_sr / (self.proc_fft_length / 2))):
+            start_flag = False
+            if (fin_threshold_point * (self.input_sr / (self.proc_fft_length / 2))) > int(self.input_sr / 2):
+                threshold_hit *= 2
+            for i in range(start_find_pos, find_range):
+                if i + 5 > find_range:
+                    break
+                # 计算连续五个采样*3 的方差，与阈值比较，判断频谱消失的位置
+                if np.var(amp_fft[i:i + 4]) < threshold_hit and \
+                   np.var(amp_fft[i + 1:i + 5]) < threshold_hit:
+                    # 定位到当前位置的前500Hz位置
+                    fin_threshold_point = i - round(forward_freq / (self.input_sr / (self.proc_fft_length / 2)))
+                    break
+            # 错误超过5把就强行钦定频率
+            loop_count += 1
+            if loop_count > 5:
+                fin_threshold_point = round(
+                    order_freq / (self.input_sr / (self.proc_fft_length / 2)))
+                break
+        return base_amp_freq, fin_threshold_point
+
+    def generate_jitter(self, input_fft, fin_threshold_point, base_amp_freq):
+        # 构造抖动
+        if fin_threshold_point <= 0:
+            return input_fft
+        for i in range(fin_threshold_point, self.proc_fft_length - fin_threshold_point):
+            # Rev.0: 调整生成概率，频率越高概率越低
+            # Rev.1: 加入幅值判定，幅度越大概率越大
+            gen_possible = abs((self.proc_fft_length / 2) - i) / ((self.proc_fft_length /
+                                                             2) - fin_threshold_point) * (base_amp_freq / 0.22)
+            if random.randint(0, 1000000) < 800000 * gen_possible:  # 0<=x<=10
+                real_value = abs(input_fft.real[i])
+                base_jitter_min = real_value * 0.5 * (1 - gen_possible)
+                base_jitter_max = real_value * 6 * gen_possible
+                amp_jitter_min = base_amp_freq * real_value * 0.5
+                amp_jitter_max = base_amp_freq * real_value * 2
+                amp_jitter_prefix = - \
+                    1 if random.randint(0, 100000) < 50000 else 1
+                jitter_prefix = - \
+                    1 if random.randint(0, 100000) < 50000 else 1
+                delta_jitter_value = random.uniform(
+                    base_jitter_min, base_jitter_max) + amp_jitter_prefix * random.uniform(amp_jitter_min, amp_jitter_max)
+                input_fft.real[
+                    i] += jitter_prefix * delta_jitter_value
+        return input_fft
 
     def save_file(self):
-        librosa.output.write_wav(self.output_file_path, self.mid_signal_carray, self.input_sr)
+        librosa.output.write_wav(self.output_file_path, self.mid_signal_carray, self.output_sr)
+
 
 if __name__ == "__main__":