Canto de voz de síntesis tutoriales ventana acoplable y el código fuente a entorno de construcción

Duplicación de Medio Ambiente canto ventana acoplable síntesis de voz

• dockerfile completa

FROM ubuntu:16.04
RUN apt-get update && \
    apt-get install -y --no-install-recommends \
        g++ \
        gcc \
        make \
        automake \
        autoconf \
        bzip2 \
        unzip \
        wget \
        sox \
        libtool \
        git \
        subversion \
        python3 \
        python3-pip \
        python3-dev \
        zlib1g-dev \
        ca-certificates \
        patch \
        ffmpeg \
	vim && \
    rm -rf /var/lib/apt/lists/*

RUN ln -s /usr/bin/python3 /usr/bin/python &&\
    ln -s /usr/bin/pip3 /usr/bin/pip

RUN pip install numpy==1.16.0 \
    scipy==1.1.0 numba==0.37.0  \
    argparse librosa audioread \
    pysoundfile picklable_itertools~=0.1.1 \
    sacred~=0.6.10 tqdm~=4.8.4 q~=2.6 \
    keras==2.1.2 tensorflow-gpu==1.8.0 \
    h5py==2.7.1 matplotlib==2.1.1 \
    pyworld -i https://pypi.douban.com/simple

$ sudo docker build -t sing:v1.0 .  # 建立镜像
$ sudo docker run -it -d -v /work:/work -p 9090:9090 --name sings sing:v1.0   # 建立容器

core Fuente

• librosa

# -*- coding:utf-8 -*-
# /usr/bin/python
'''
@Describe:  音频处理
@Evn     :  pip install librosa
@Date    :  2019-08-04  15:23
'''

'''
pip install librosa
ffmpeg is very stronger.

librosa.beat:用于检测速度和节拍
librosa.core:用于从磁盘加载音频和计算各种频谱图
librosa.decompose:实现矩阵分解进行谐波和冲击源分离通用频谱图分解
librosa.display:音频特征的显示
librosa.effects:时域音频处理，音高变换和时间拉伸，时域包装器。
librosa.feature:特征提取和操作：色度图，伪常数Q（对数频率）变换，Mel频谱图，MFCC和调谐估计
librosa.filters:滤波器生成色度。伪CQT、CQT等
librosa.onset:其实检测和起始强度计算。
librosa.segment:用于结构分段的函数
librosa.swquence:顺序建模功能
librosa.util:辅助工具（规范化。填充、居中）
'''
# Beat tracking example
#from __future__ import print_function
import librosa
from librosa import display
import matplotlib.pyplot as plt
# 1. Get the file path to the included audio example
filename = librosa.util.example_audio_file()

# 2. Load the audio as a waveform `y`data，sampling_rate
#    Store the sampling rate as `sr`
filename = "a.wav"
y, sr = librosa.load(filename)
plt.figure(figsize=(12, 4))
display.waveplot(y,sr=sr)
plt.show()
plt.savefig('./test.png')

print("y",y,"\nsr",sr)

# 3. Run the default beat tracker
tempo, beat_frames = librosa.beat.beat_track(y=y, sr=sr)
print("tempo",tempo, "\nbeat_frames",beat_frames)

print('Estimated tempo: {:.2f} beats per minute'.format(tempo))

# 4. Convert the frame indices of beat events into timestamps
beat_times = librosa.frames_to_time(beat_frames, sr=sr)

# 5.Separate harmonics and percussives into two waveforms
y_harmonic, y_percussive = librosa.effects.hpss(y)
print(y_harmonic,'\n', y_percussive)

• Mdmom

# -*- coding:utf-8 -*-
# /usr/bin/python
'''
@Describe: madmom
@Evn     :  pip install madmom
@Date    :  2019-09-01  12:57
'''

import numpy as np
import matplotlib.pyplot as plt

import madmom
filename = "../datasets/test.wav"
signal, sample_rate = madmom.audio.signal.load_wave_file(filename)
print('signal',signal,'sample_rate' ,sample_rate)
sig = madmom.audio.signal.Signal(filename)
sample_rate = sig.sample_rate
print('sig',sig,'sample_rate',sample_rate)

fs = madmom.audio.signal.FramedSignal(sig, frame_size=2048, hop_size=441)
print('fs',fs)
print(fs.frame_rate, fs.hop_size, fs[0])

stft = madmom.audio.stft.STFT(fs)
print(stft[0:2])
spec = madmom.audio.spectrogram.Spectrogram(stft)
plt.figure()
plt.imshow(spec[:, :200].T, aspect='auto', origin='lower')
plt.savefig('./test.png')

# calculate the difference
diff = np.diff(spec, axis=0)
# keep only the positive differences
pos_diff = np.maximum(0, diff)
# sum everything to get the spectral flux
sf = np.sum(pos_diff, axis=1)

plt.figure()
plt.imshow(spec[:, :200].T, origin='lower', aspect='auto')
plt.savefig('./test1.png')

plt.figure()
plt.imshow(pos_diff[:, :200].T, origin='lower', aspect='auto')
plt.savefig('./test2.png')

plt.figure()
plt.plot(sf)
plt.savefig('./test3.png')

filt_spec = madmom.audio.spectrogram.FilteredSpectrogram(spec, filterbank=madmom.audio.filters.LogFilterbank,
                                                         num_bands=24)
plt.imshow(filt_spec.T, origin='lower', aspect='auto')
plt.figure()
log_spec = madmom.audio.spectrogram.LogarithmicSpectrogram(filt_spec, add=1)
plt.imshow(log_spec.T, origin='lower', aspect='auto')
plt.savefig('./test4.png')

Generando muestras de voz

• Link: https: //pan.baidu.com/s/1Tl9mSQM2kmD8vgBfKVvFaA Contraseña: ma2r

Ver los detalles de la columna inmediatamente desbloquear todas las columnas