Dan's Update
- New classes for noisegate & audio stream - Ability to manipulate raw data in real-time
This commit is contained in:
Logan Cusano
151
NoiseGatev2.py
151
NoiseGatev2.py
@@ -1,25 +1,24 @@
|
||||
import audioop
|
||||
import math
|
||||
import pyaudio
|
||||
import struct
|
||||
import discord
|
||||
import numpy
|
||||
import time
|
||||
from threading import Thread
|
||||
|
||||
sound_buffer = []
|
||||
THRESHOLD = 50
|
||||
voice_connection = None
|
||||
|
||||
|
||||
class AudioStream(Thread):
|
||||
def __init__(self, _channels: int = 1, _sample_rate: int = 48000, _frames_per_buffer: int = 2048,
|
||||
class AudioStream:
|
||||
def __init__(self, _channels: int = 2, _sample_rate: int = 48000, _frames_per_buffer: int = 1024,
|
||||
_input_device_index: int = None, _output_device_index: int = None, _input: bool = True,
|
||||
_output: bool = True):
|
||||
super(AudioStream, self).__init__()
|
||||
_output: bool = True, _init_on_startup: bool = True):
|
||||
self.paInstance_kwargs = {
|
||||
'format': pyaudio.paFloat32,
|
||||
'format': pyaudio.paInt16,
|
||||
'channels': _channels,
|
||||
'rate': _sample_rate,
|
||||
'input': _input,
|
||||
'output': _output,
|
||||
'frames_per_buffer': _frames_per_buffer,
|
||||
'stream_callback': callback
|
||||
'frames_per_buffer': _frames_per_buffer
|
||||
}
|
||||
|
||||
if _input_device_index:
|
||||
@@ -36,15 +35,18 @@ class AudioStream(Thread):
|
||||
print(f"[AudioStream.__init__]:\tOutput was not enabled."
|
||||
f" Reinitialize with '_output=True'")
|
||||
|
||||
# Init PyAudio instance
|
||||
self.paInstance = pyaudio.PyAudio()
|
||||
if _init_on_startup:
|
||||
# Init PyAudio instance
|
||||
print("Creating PyAudio instance")
|
||||
self.paInstance = pyaudio.PyAudio()
|
||||
|
||||
# Define and initialize stream object if we have been passed a device ID (pyaudio.open)
|
||||
self.stream = None
|
||||
if _output_device_index and _input_device_index:
|
||||
self.init_stream()
|
||||
# Define and initialize stream object if we have been passed a device ID (pyaudio.open)
|
||||
self.stream = None
|
||||
|
||||
# temp section
|
||||
if _output_device_index or _input_device_index:
|
||||
if _init_on_startup:
|
||||
print("Init stream")
|
||||
self.init_stream()
|
||||
|
||||
def init_stream(self, _new_output_device_index: int = None, _new_input_device_index: int = None):
|
||||
# Check what device was asked to be changed (or set)
|
||||
@@ -64,7 +66,7 @@ class AudioStream(Thread):
|
||||
|
||||
self.close_if_open()
|
||||
|
||||
# Reopen the stream
|
||||
# Open the stream
|
||||
self.stream = self.paInstance.open(**self.paInstance_kwargs)
|
||||
|
||||
def close_if_open(self):
|
||||
@@ -75,56 +77,109 @@ class AudioStream(Thread):
|
||||
self.stream.close()
|
||||
print(f"[ReopenStream.close_if_open]:\t Stream was open; It was closed.")
|
||||
|
||||
def list_devices(self, _show_input_devices: bool = True, _show_output_devices: bool = True):
|
||||
def list_devices(self, _display_input_devices: bool = True, _display_output_devices: bool = True):
|
||||
info = self.paInstance.get_host_api_info_by_index(0)
|
||||
numdevices = info.get('deviceCount')
|
||||
|
||||
devices = {
|
||||
'Input': {},
|
||||
'Output': {}
|
||||
}
|
||||
for i in range(0, numdevices):
|
||||
if (self.paInstance.get_device_info_by_host_api_device_index(0, i).get('maxInputChannels')) > 0:
|
||||
if _show_input_devices:
|
||||
print("Input Device id ", i, " - ",
|
||||
self.paInstance.get_device_info_by_host_api_device_index(0, i).get('name'))
|
||||
input_device = self.paInstance.get_device_info_by_host_api_device_index(0, i).get('name')
|
||||
devices['Input'][i] = input_device
|
||||
if _display_input_devices:
|
||||
print("Input Device id ", i, " - ", input_device)
|
||||
|
||||
if (self.paInstance.get_device_info_by_host_api_device_index(0, i).get('maxOutputChannels')) > 0:
|
||||
if _show_output_devices:
|
||||
print("Output Device id ", i, " - ",
|
||||
self.paInstance.get_device_info_by_host_api_device_index(0, i).get('name'))
|
||||
output_device = self.paInstance.get_device_info_by_host_api_device_index(0, i).get('name')
|
||||
devices['Output'][i] = output_device
|
||||
if _display_output_devices:
|
||||
print("Output Device id ", i, " - ", output_device)
|
||||
|
||||
return devices
|
||||
|
||||
class NoiseGate(AudioStream):
|
||||
def __init__(self, **kwargs):
|
||||
super().__init__(**kwargs)
|
||||
|
||||
def run(self) -> None:
|
||||
# Start the audio stream
|
||||
self.stream.start_stream()
|
||||
|
||||
global sound_buffer
|
||||
# While the stream is running, display the stream buffer in floats?
|
||||
while self.stream.is_active():
|
||||
if len(sound_buffer) > 0:
|
||||
for buffer in sound_buffer:
|
||||
volume_normalization = numpy.linalg.norm(numpy.fromstring(buffer)) * 10
|
||||
print(str(float(volume_normalization)))
|
||||
|
||||
self.stream.stop_stream()
|
||||
async def stop(self):
|
||||
await voice_connection.disconnect()
|
||||
self.close_if_open()
|
||||
self.stream.close()
|
||||
|
||||
self.paInstance.terminate()
|
||||
|
||||
|
||||
def callback(in_data, frame_count, time_info, status):
|
||||
global sound_buffer
|
||||
class NoiseGate(AudioStream):
|
||||
def __init__(self, _voice_connection, **kwargs):
|
||||
super(NoiseGate, self).__init__(_init_on_startup=True, **kwargs)
|
||||
global voice_connection
|
||||
voice_connection = _voice_connection
|
||||
self.NGStream = NoiseGateStream(self)
|
||||
|
||||
sound_buffer.append(in_data)
|
||||
def run(self) -> None:
|
||||
global voice_connection
|
||||
# Start the audio stream
|
||||
self.stream.start_stream()
|
||||
voice_connection.play(self.NGStream)
|
||||
|
||||
return in_data, pyaudio.paContinue
|
||||
async def close(self):
|
||||
await voice_connection.disconnect()
|
||||
if self.stream.is_active:
|
||||
self.stream.stop_stream()
|
||||
|
||||
class NoiseGateStream(discord.AudioSource):
|
||||
def __init__(self, _stream):
|
||||
super(NoiseGateStream, self).__init__()
|
||||
self.stream = _stream # The actual audio stream object
|
||||
self.NG_fadeout = 480/20 # Fadeout value used to hold the noisegate after de-triggering
|
||||
self.NG_fadeout_count = 0 # A count set when the noisegate is triggered and was de-triggered
|
||||
self.process_set_count = 0 # Counts how many processes have been made in order to limit the prints
|
||||
|
||||
def read(self):
|
||||
try:
|
||||
while voice_connection.is_connected():
|
||||
curr_buffer = bytearray(self.stream.stream.read(960))
|
||||
buffer_rms = audioop.rms(curr_buffer, 2)
|
||||
if buffer_rms > 0:
|
||||
buffer_decibel = 20 * math.log10(buffer_rms)
|
||||
|
||||
if self.process_set_count % 25 == 0:
|
||||
print(f"{buffer_decibel} db")
|
||||
|
||||
if buffer_decibel >= THRESHOLD:
|
||||
self.NG_fadeout_count = self.NG_fadeout
|
||||
self.process_set_count += 1
|
||||
|
||||
return bytes(curr_buffer)
|
||||
|
||||
else:
|
||||
if self.NG_fadeout_count > 0:
|
||||
self.NG_fadeout_count -= 1
|
||||
print(f"Frames in fadeout remaining: {self.NG_fadeout_count}")
|
||||
self.process_set_count += 1
|
||||
|
||||
return bytes(curr_buffer)
|
||||
|
||||
except OSError as e:
|
||||
pass
|
||||
|
||||
def audio_datalist_set_volume(self, datalist, volume):
|
||||
""" Change value of list of audio chunks """
|
||||
sound_level = (volume / 100.)
|
||||
|
||||
for i in range(len(datalist)):
|
||||
chunk = numpy.fromstring(datalist[i], numpy.int16)
|
||||
|
||||
chunk = chunk * sound_level
|
||||
|
||||
datalist[i] = chunk.astype(numpy.int16)
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
|
||||
input_index = int(input("Input:\t"))
|
||||
output_index = int(input("Output:\t"))
|
||||
|
||||
ng = NoiseGate(_input_device_index=input_index, _output_device_index=output_index)
|
||||
|
||||
ng.list_devices()
|
||||
|
||||
ng.start()
|
||||
|
||||
Reference in New Issue
Block a user