使用tensorflow.contrib.signal重建信号会导致放大或调制(帧,overlap_and_add,stft等)

时间:2018-01-29 21:24:00

标签: python tensorflow time-series signal-processing

更新 :我已经在librosa中对此进行了重新实现以进行比较,结果确实与tensorflow的结果非常不同。 Librosa给出了我期望的结果(但不是张量流)。

我已将此作为issue发布在tensorflow回购中,但它在那里很安静所以我在这里尝试。此外,我不确定它是否是张量流中的错误或代表我的用户错误。为了完整起见,我也将在此处包含完整的来源和结果。

A。)当我使用hann窗口(也尝试了汉明)从frame_length=1024frame_step=256(即25%跳跃大小,75%重叠)的信号创建帧时,然后我重建使用overlap_and_add,我希望信号能够正确重建(因为COLA等)。但相反,它的幅度恰好是两倍。我需要将结果信号除以2才能正确。

B。)如果我使用STFT创建一系列重叠频谱图,然后使用反向STFT重建,再次使用frame_length=1024frame_step=256,再次以双幅度重建信号。

我意识到为什么会出现这种情况(hann在50%重叠时的单位增益,因此75%的重叠将使信号加倍)。但重建功能考虑到这一点是不正常的吗?例如。 librosa istft确实返回具有正确幅度的信号,而tensorflow返回两倍。

下进行。) 在任何其他frame_step处,存在严重的幅度调制。见下图。这看起来并不正确。

更新:如果我在window_fn=tf.contrib.signal.inverse_stft_window_fn(frame_step)中明确设置inverse_stft,则输出正确无误。因此,frame_step中的inverse_stft似乎没有被传递到窗口函数中(这也是结果提示的内容)。

原始数据:

enter image description here

帧的tensorflow输出+ overlap_and_add:

enter image description here

enter image description here

enter image description here

enter image description here

stft + istft的张量流输出:

enter image description here

enter image description here

enter image description here

enter image description here

stft + istft的librosa输出:

enter image description here

enter image description here

enter image description here

enter image description here

tensorflow代码:

from __future__ import print_function
from __future__ import division

import numpy as np
import scipy.io.wavfile
import math
import random
import matplotlib.pyplot as plt

import tensorflow as tf
out_prefix = 'tensorflow'


def plot(data, title, do_save=True):
    plt.figure(figsize=(20,5))
    plt.plot(data[:3*frame_length])
    plt.ylim([-1, 1])
    plt.title(title)
    plt.grid()
    if do_save: plt.savefig(title + '.png')
    plt.show()


def reconstruct_from_frames(x, frame_length, frame_step):
    name = 'frame'
    frames_T = tf.contrib.signal.frame(x, frame_length=frame_length, frame_step=frame_step)
    windowed_frames_T = frames_T * tf.contrib.signal.hann_window(frame_length, periodic=True)
    output_T = tf.contrib.signal.overlap_and_add(windowed_frames_T, frame_step=frame_step)
    return name, output_T


def reconstruct_from_stft(x, frame_length, frame_step):
    name = 'stft'
    spectrograms_T = tf.contrib.signal.stft(x, frame_length, frame_step)
    output_T = tf.contrib.signal.inverse_stft(spectrograms_T, frame_length, frame_step)
    return name, output_T


def test(fn, input_data):
    print('-'*80)
    tf.reset_default_graph()
    input_T = tf.placeholder(tf.float32, [None]) 
    name, output_T = fn(input_T, frame_length, frame_step)

    title = "{}.{}.{}.l{}.s{}".format(out_prefix, sample_rate, name, frame_length, frame_step)
    print(title)

    with tf.Session():
        output_data =  output_T.eval({input_T:input_data})

#    output_data /= frame_length/frame_step/2 # tensorflow needs this to normalise amp
    plot(output_data, title)
    scipy.io.wavfile.write(title+'.wav', sample_rate, output_data)


def generate_data(duration_secs, sample_rate, num_sin, min_freq=10, max_freq=500, rnd_seed=0, max_val=0):
    '''generate signal from multiple random sin waves'''
    if rnd_seed>0: random.seed(rnd_seed)
    data = np.zeros([duration_secs*sample_rate], np.float32)
    for i in range(num_sin):
        w = np.float32(np.sin(np.linspace(0, math.pi*2*random.randrange(min_freq, max_freq), num=duration_secs*sample_rate)))
        data += random.random() * w
    if max_val>0:
        data *= max_val / np.max(np.abs(data))
    return data


frame_length = 1024
sample_rate = 22050

input_data = generate_data(duration_secs=1, sample_rate=sample_rate, num_sin=1, rnd_seed=2, max_val=0.5)

title = "{}.orig".format(sample_rate)
plot(input_data, title)
scipy.io.wavfile.write(title+'.wav', sample_rate, input_data)

for frame_step in [256, 512, 768, 1024]:
    test(reconstruct_from_frames, input_data)
    test(reconstruct_from_stft, input_data)

print('done.')

librosa代码:

from __future__ import print_function
from __future__ import division

import numpy as np
import scipy.io.wavfile
import math
import random
import matplotlib.pyplot as plt

import librosa.core as lc
out_prefix = 'librosa'


def plot(data, title, do_save=True):
    plt.figure(figsize=(20,5))
    plt.plot(data[:3*frame_length])
    plt.ylim([-1, 1])
    plt.title(title)
    plt.grid()
    if do_save: plt.savefig(title + '.png')
    plt.show()


def reconstruct_from_stft(x, frame_length, frame_step):
    name = 'stft'
    stft = lc.stft(x, n_fft=frame_length, hop_length=frame_step)
    istft = lc.istft(stft, frame_step)
    return name, istft


def test(fn, input_data):
    print('-'*80)
    name, output_data = fn(input_data, frame_length, frame_step)

    title = "{}.{}.{}.l{}.s{}".format(out_prefix, sample_rate, name, frame_length, frame_step)
    print(title)

#    output_data /= frame_length/frame_step/2 # tensorflow needs this to normalise amp
    plot(output_data, title)
    scipy.io.wavfile.write(title+'.wav', sample_rate, output_data)


def generate_data(duration_secs, sample_rate, num_sin, min_freq=10, max_freq=500, rnd_seed=0, max_val=0):
    '''generate signal from multiple random sin waves'''
    if rnd_seed>0: random.seed(rnd_seed)
    data = np.zeros([duration_secs*sample_rate], np.float32)
    for i in range(num_sin):
        w = np.float32(np.sin(np.linspace(0, math.pi*2*random.randrange(min_freq, max_freq), num=duration_secs*sample_rate)))
        data += random.random() * w
    if max_val>0:
        data *= max_val / np.max(np.abs(data))
    return data


frame_length = 1024
sample_rate = 22050

input_data = generate_data(duration_secs=1, sample_rate=sample_rate, num_sin=1, rnd_seed=2, max_val=0.5)

title = "{}.orig".format(sample_rate)
plot(input_data, title)
scipy.io.wavfile.write(title+'.wav', sample_rate, input_data)

for frame_step in [256, 512, 768, 1024]:
    test(reconstruct_from_stft, input_data)

print('done.')
  • Linux Ubuntu 16.04
  • 从二进制文件v1.4.0-19-ga52c8d9,1.4.1
  • 安装的Tensorflow
  • Python 2.7.14 | Anaconda定制(64位)| (默认,2017年10月16日,17:29:19)。 IPython 5.4.1
  • Cuda版本8.0,V8.0.61,cuDNN 6
  • Geforce GTX 970M,驱动程序版本:384.111

(刚试过TF1.5,Cuda9.0,cuDNN 7.0.5,结果相同)。

2 个答案:

答案 0 :(得分:0)

无限序列的Von Hann窗口,50%的重叠总和为平坦的单位增益。如果重叠率为25%,则每单位时间的窗口数量会翻倍,从而使增益加倍。

重叠 - 添加快速卷积滤波通常在没有重叠和没有(非矩形)窗口的情况下完成,只是至少足够的零填充至少是滤波器函数的脉冲响应的长度。任何重叠百分比都不应包括任何添加的零填充长度。

答案 1 :(得分:0)

您应该使用tf.signal.inverse_stft_window_fn

window_fn=tf.signal.inverse_stft_window_fn(frame_step)
tf_istfts=tf.signal.inverse_stft(tf_stfts, frame_length=frame_length, frame_step=frame_step, fft_length=fft_length, window_fn=window_fn)}

inverse_stft_window_fn上查看更多信息