|
|
|
|
|
|
|
|
from typing import List |
|
|
|
|
|
import matplotlib.pyplot as plt |
|
|
import numpy as np |
|
|
|
|
|
|
|
|
def process_speech_probs(signal: np.ndarray, speech_probs: List[float], frame_step: int) -> np.ndarray: |
|
|
speech_probs_ = list() |
|
|
for p in speech_probs[1:]: |
|
|
speech_probs_.extend([p] * frame_step) |
|
|
|
|
|
pad = (signal.shape[0] - len(speech_probs_)) |
|
|
speech_probs_ = speech_probs_ + [0.0] * pad |
|
|
speech_probs_ = np.array(speech_probs_, dtype=np.float32) |
|
|
|
|
|
if len(speech_probs_) != len(signal): |
|
|
raise AssertionError |
|
|
return speech_probs_ |
|
|
|
|
|
|
|
|
def make_visualization(signal: np.ndarray, speech_probs, sample_rate: int): |
|
|
time = np.arange(0, len(signal)) / sample_rate |
|
|
plt.figure(figsize=(12, 5)) |
|
|
plt.plot(time, signal, color='b') |
|
|
plt.plot(time, speech_probs, color='gray') |
|
|
plt.show() |
|
|
return |
|
|
|
|
|
|
|
|
if __name__ == "__main__": |
|
|
pass |
|
|
|
