Data indices as result of vad instead of time

[rhythmize]

How can I get data incides where speech starts or ends, instead of time when the same happens?

[Sadam1195]

import sounddevice as sd
import numpy as np
import soundfile as sf
from vad import VoiceActivityDetector

filename = '/home/sadam/saddam.wav' # path to your filename here
y, sr = sf.read(filename)

def get_chunk(y, sr, from_ms, to_ms):
    from_sample = np.round(from_ms*(sr)).astype(int)
    to_sample = np.round(to_ms*(sr)).astype(int)
    return y[from_sample:to_sample]


voice = VoiceActivityDetector(filename)
d_sp = voice.detect_speech()
rd_lb = v.convert_windows_to_readible_labels(d_sp)
conct_data = [] # this will store all speech parts
for i in range(len(rd_lb)):  
    data = get_chunk(y,sr, rd_lb[i]['speech_begin'],rd_lb[i]['speech_end'])
    conct_data.extend(data)

sd.play(conct_data,sr)
sd.wait()