实时音频音调转换器Python

问题描述

我正在尝试使用 wav 文件制作实时变调器

我从 https://github.com/waynesun626/Real-Time-Pitch-Shifter/blob/main/shifter.py 获得了代码

并修改此代码以在播放时更改 wav 文件的音调。

import numpy as np
import sox
import pyaudio,struct,wave
import tkinter
import pygame

# sample rate in Hz
RATE = 44100
RECORD_SECONDS = 5
BLOCKLEN=800

MAXVALUE = 2**15-1

k=0

# Number of blocks to run for
num_blocks = int(RATE / BLOCKLEN * RECORD_SECONDS)

p = pyaudio.PyAudio()
wf = wave.open("sample.wav",'rb')
PA_FORMAT = pyaudio.paInt16

stream = p.open(format=p.get_format_from_width(wf.getsampwidth()),channels=2,rate=44100,output=True,frames_per_buffer = 800)

# stream = p.open(
#         format      = PA_FORMAT,#         channels    = 1,#         rate        = RATE,#         input       = True,#         output      = True,#         frames_per_buffer = 800)

CONTINUE = True
KEYPRESS = False

def my_function(event):
    global CONTINUE
    global KEYPRESS
    global k
    print('You pressed ' + event.char)
    if event.char == 'q':
      print('Good bye')
      CONTINUE = False
    if event .char=='+':
        k += 1
    if event .char=='-':
        k -= 1
    if event .char=='=':
        k = 0
    KEYPRESS = True

def play():
    pygame.init()
    pygame.mixer.init()
    pygame.mixer.music.load('aroha.wav')
    pygame.mixer.music.play()


root = tkinter.Tk()
root.bind("<Key>",my_function)

button1 = tkinter.Button(text='play')
button1.config(command = play)
print('Press keys for sound.')
print('Press "q" to quit')
button1.pack()


# Start loopfgh
while CONTINUE:
    root.update()

    # Get frames from audio input stream
    # input_bytes = stream.read(BLOCKLEN)       # BLOCKLEN = number of frames read
    input_bytes = stream.read(BLOCKLEN,exception_on_overflow = False)   # BLOCKLEN = number of frames read
    # print(input_bytes)
    # Convert binary data to tuple of numbers
    input_tuple = struct.unpack('h' * BLOCKLEN,input_bytes)
    data=np.array(input_tuple)

    scaled = np.frombuffer(input_bytes,np.int16)

    # create a transformer
    tfm = sox.Transformer()


    # shift the pitch up by 1 semitones
    tfm.pitch(k)

    y_out=2*tfm.build_array(input_array=scaled,sample_rate_in=RATE)

    y = np.clip(y_out.astype(int),-MAXVALUE,MAXVALUE)     # Clipping

    output_bytes = struct.pack('h' * BLOCKLEN,*y)

    # Write binary data to audio output stream
    stream.write(output_bytes,BLOCKLEN)

print('* Finished')

stream.stop_stream()
stream.close()
p.terminate()

注释掉的“流”是原始的。

到目前为止，我发现的问题是渠道不同。

由于 wav 文件的通道是两次，因此 input_bytes（第 94 行）从 1600 字节变为 3200 字节。

因此，这会导致 input_tuple（第 97 行）出现问题，因为 struct.unpack 仅适用于 1600 个字节。

我认为应该还有更多问题

请帮忙！

我很绝望。我需要你的帮助！

解决方法

暂无找到可以解决该程序问题的有效方法，小编努力寻找整理中！

如果你已经找到好的解决方法，欢迎将解决方案带上本链接一起发送给小编。

小编邮箱:dio#foxmail.com (将#修改为@）