Merge pull request #39 from wst24365888/dev

chore: remove unused dependency

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "voice-presentation-control"
-version = "1.0.1"
+version = "1.0.2"
 description = "voice-presentation-control is a tool that allows you to control your presentation using voice when you don't have a presentation pen or when it's inconvinient to use the keyboard."
 authors = ["Xyphuz <[email protected]>"]
 readme = "README.md"
@@ -14,14 +14,12 @@ include = [
 ]
 
 [tool.poetry.dependencies]
-python = ">=3.9,<3.11"
+python = "^3.9"
 PyAudio = "^0.2.11"
 typer = "^0.4.1"
 PyAutoGUI = "^0.9.53"
 numpy = "^1.22.3"
 vosk = "^0.3.32"
-scipy = "^1.8.1"
-logmmse = "^1.5"
 
 [tool.poetry.dev-dependencies]
 pytest = "^7.1.2"

voice_presentation_control/__init__.py

@@ -1,2 +1,2 @@
 __app_name__ = "voice_presentation_control"
-__version__ = "1.0.1"
+__version__ = "1.0.2"

voice_presentation_control/controller.py

@@ -7,7 +7,6 @@
 from queue import Queue
 from typing import List
 
-import logmmse
 import numpy as np
 import pyaudio
 
@@ -162,11 +161,6 @@ def volume_process(self, wave_values: np.ndarray) -> array:
 
         return wave_values_arr
 
-    def denoise(self, wave_values: np.ndarray) -> np.ndarray:
-        wave_values = logmmse.logmmse(data=wave_values, sampling_rate=self.rate, noise_threshold=0.15, window_size=0)
-
-        return wave_values
-
     def freqs_process(self, fft_wave: np.ndarray) -> np.ndarray:
         sample_num = len(fft_wave)
 
@@ -176,14 +170,22 @@ def freqs_process(self, fft_wave: np.ndarray) -> np.ndarray:
 
         # filter
         fft_filter = vib_fft.copy()
-        noise_indices = np.where((abs(fft_freqs) > 8000) & (abs(fft_freqs) < 10000))
-        fft_filter[noise_indices] = fft_filter[noise_indices] * 0.5  # .1
+        fft_filter = fft_filter / 10
+
+        noise_indices = np.where(((abs(fft_freqs) >= 100) & (abs(fft_freqs) < 200)))  # n
+        fft_filter[noise_indices] = fft_filter[noise_indices] * 10
+
+        noise_indices = np.where(((abs(fft_freqs) >= 1500) & (abs(fft_freqs) < 2000)))  # p
+        fft_filter[noise_indices] = fft_filter[noise_indices] * 2
+
+        noise_indices = np.where(((abs(fft_freqs) >= 3000) & (abs(fft_freqs) < 3500)))  # n
+        fft_filter[noise_indices] = fft_filter[noise_indices] * 5
 
-        noise_indices = np.where(abs(fft_freqs) >= 10000)
-        fft_filter[noise_indices] = fft_filter[noise_indices] * 0.1  # .05
+        noise_indices = np.where(((abs(fft_freqs) >= 5000) & (abs(fft_freqs) < 8000)))  # x
+        fft_filter[noise_indices] = fft_filter[noise_indices]
 
-        noise_indices = np.where(abs(fft_freqs) >= 15000)
-        fft_filter[noise_indices] = fft_filter[noise_indices] * 0
+        noise_indices = np.where((abs(fft_freqs) > 8000))
+        fft_filter[noise_indices] = fft_filter[noise_indices] * 0  # .1
 
         filter_wave_ifft = np.fft.ifft(fft_filter).real
 
@@ -193,7 +195,6 @@ def audio_preprocess(self, record_frames: List[bytes]) -> bytes:
         wave_values = np.array(array("h", b"".join(record_frames)))
 
         wave_values = self.freqs_process(wave_values)
-        # wave_values = self.denoise(wave_values)
         wave_values_arr = self.volume_process(wave_values)
 
         return wave_values_arr.tobytes()