import numpy as np
from itertools import cycle, product
from malaya_speech.model.frame import Frame
from herpetologist import check_type
from typing import List, Tuple
from itertools import groupby
def get_ax(
ax=None,
xlim=(0, 1000),
ylim=(0, 1),
yaxis=False,
time=True,
**kwargs
):
try:
import seaborn as sns
import matplotlib.pyplot as plt
except BaseException:
raise ValueError(
'seaborn and matplotlib not installed. Please install it by `pip install matplotlib seaborn` and try again.'
)
if ax is None:
ax = plt.gca()
ax.set_xlim(xlim)
if time:
ax.set_xlabel('Time')
else:
ax.set_xticklabels([])
ax.set_ylim(ylim)
ax.axes.get_yaxis().set_visible(yaxis)
return ax
def get_styles(size):
try:
from matplotlib.cm import get_cmap
except BaseException:
raise ValueError(
'matplotlib not installed. Please install it by `pip install matplotlib` and try again.'
)
linewidth = [3, 1]
linestyle = ['solid', 'dashed', 'dotted']
cm = get_cmap('Set1')
colors = [cm(1.0 * i / 8) for i in range(9)]
style_generator = cycle(product(linestyle, linewidth, colors))
styles = [next(style_generator) for _ in range(size)]
return styles
[docs]def visualize_vad(
signal,
preds: List[Tuple[Frame, bool]],
sample_rate: int = 16000,
figsize: Tuple[int, int] = (15, 3),
ax=None,
**kwargs
):
"""
Visualize signal given VAD labels. Green means got voice activity, while Red is not.
Parameters
-----------
signal: list / np.array
preds: List[Tuple[Frame, bool]]
sample_rate: int, optional (default=16000)
figsize: Tuple[int, int], optional (default=(15, 7))
matplotlib figure size.
"""
try:
import seaborn as sns
import matplotlib.pyplot as plt
except BaseException:
raise ValueError(
'seaborn and matplotlib not installed. Please install it by `pip install matplotlib seaborn` and try again.'
)
if ax is None:
sns.set()
fig = plt.figure(figsize=figsize)
ax = fig.add_subplot(1, 1, 1)
plot = True
else:
min_timestamp = min([i[0].timestamp for i in preds])
max_timestamp = max([i[0].timestamp + i[0].duration for i in preds])
ax = get_ax(
ax,
xlim=(min_timestamp, max_timestamp),
ylim=(np.min(signal), np.max(signal)),
**kwargs
)
plot = False
ax.plot([i / sample_rate for i in range(len(signal))], signal)
for predictions in preds:
color = 'g' if predictions[1] else 'r'
p = predictions[0]
ax.axvspan(
p.timestamp, p.timestamp + p.duration, alpha=0.5, color=color
)
if plot:
plt.xlabel('Time (s)', size=20)
plt.ylabel('Amplitude', size=20)
plt.xticks(size=15)
plt.yticks(size=15)
plt.show()
[docs]def plot_classification(
preds,
description,
ax=None,
fontsize_text=14,
x_text=0.05,
y_text=0.2,
ylim=(0.1, 0.9),
figsize: Tuple[int, int] = (15, 3),
**kwargs
):
"""
Visualize probability / boolean.
Parameters
-----------
preds: List[Tuple[Frame, label]]
description: str
ax: ax, optional (default = None)
fontsize_text: int, optional (default = 14)
x_text: float, optional (default = 0.05)
y_text: float, optional (default = 0.2)
"""
try:
import seaborn as sns
import matplotlib.pyplot as plt
except BaseException:
raise ValueError(
'seaborn and matplotlib not installed. Please install it by `pip install matplotlib seaborn` and try again.'
)
if ax is None:
fig = plt.figure(figsize=figsize)
ax = fig.add_subplot(1, 1, 1)
if isinstance(preds[0][1], float) or isinstance(preds[0][1], np.float32):
hline = False
else:
hline = True
min_timestamp = min([i[0].timestamp for i in preds])
max_timestamp = max([i[0].timestamp + i[0].duration for i in preds])
ax = get_ax(ax, xlim=(min_timestamp, max_timestamp), **kwargs)
if hline:
x = [i[1] for i in preds]
labels = sorted(list(set(x)))
styles = get_styles(len(labels))
styles = {label: style for label, style in zip(labels, styles)}
xs = [labels.index(i[1]) for i in preds]
a = np.array(xs)
std = (a - np.min(a)) / (np.max(a) - np.min(a))
scaled = std * (ylim[1] - ylim[0]) + ylim[0]
for i in range(len(preds)):
linestyle, linewidth, color = styles[x[i]]
ax.hlines(
scaled[i],
preds[i][0].timestamp,
preds[i][0].timestamp + preds[i][0].duration,
color,
linewidth=linewidth,
linestyle=linestyle,
label=x[i],
)
ax.vlines(
preds[i][0].timestamp,
scaled[i] + 0.05,
scaled[i] - 0.05,
color,
linewidth=1,
linestyle='solid',
)
ax.vlines(
preds[i][0].timestamp + preds[i][0].duration,
scaled[i] + 0.05,
scaled[i] - 0.05,
color,
linewidth=1,
linestyle='solid',
)
H, L = ax.get_legend_handles_labels()
HL = groupby(
sorted(zip(H, L), key=lambda h_l: h_l[1]),
key=lambda h_l: h_l[1],
)
H, L = zip(*list((next(h_l)[0], l) for l, h_l in HL))
ax.legend(
H,
L,
bbox_to_anchor=(0, 1),
loc=3,
ncol=5,
borderaxespad=0.0,
frameon=False,
)
else:
x = [i[0].timestamp for i in preds]
y = [i[1] for i in preds]
ax.plot(x, y)
x = [i[0].timestamp for i in preds]
ax.text(
x[int(len(x) * x_text)], y_text, description, fontsize=fontsize_text
)
return ax
