File size: 5,826 Bytes
1ac9078
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
import torch

# This is copied from silero-vad's vad_utils.py:
# https://github.com/snakers4/silero-vad/blob/f6b1294cb27590fb2452899df98fb234dfef1134/utils_vad.py#L340
# (except changed defaults)

# Their licence is MIT, same as ours: https://github.com/snakers4/silero-vad/blob/f6b1294cb27590fb2452899df98fb234dfef1134/LICENSE


class VADIterator:
    def __init__(
        self,
        model,
        threshold: float = 0.5,
        sampling_rate: int = 16000,
        min_silence_duration_ms: int = 500,  # makes sense on one recording that I checked
        speech_pad_ms: int = 100,  # same
    ):
        """
        Class for stream imitation

        Parameters
        ----------
        model: preloaded .jit silero VAD model

        threshold: float (default - 0.5)
            Speech threshold. Silero VAD outputs speech probabilities for each audio chunk, probabilities ABOVE this value are considered as SPEECH.
            It is better to tune this parameter for each dataset separately, but "lazy" 0.5 is pretty good for most datasets.

        sampling_rate: int (default - 16000)
            Currently silero VAD models support 8000 and 16000 sample rates

        min_silence_duration_ms: int (default - 100 milliseconds)
            In the end of each speech chunk wait for min_silence_duration_ms before separating it

        speech_pad_ms: int (default - 30 milliseconds)
            Final speech chunks are padded by speech_pad_ms each side
        """

        self.model = model
        self.threshold = threshold
        self.sampling_rate = sampling_rate

        if sampling_rate not in [8000, 16000]:
            raise ValueError(
                "VADIterator does not support sampling rates other than [8000, 16000]"
            )

        self.min_silence_samples = sampling_rate * min_silence_duration_ms / 1000
        self.speech_pad_samples = sampling_rate * speech_pad_ms / 1000
        self.reset_states()

    def reset_states(self):

        self.model.reset_states()
        self.triggered = False
        self.temp_end = 0
        self.current_sample = 0

    def __call__(self, x, return_seconds=False):
        """
        x: torch.Tensor
            audio chunk (see examples in repo)

        return_seconds: bool (default - False)
            whether return timestamps in seconds (default - samples)
        """

        if not torch.is_tensor(x):
            try:
                x = torch.Tensor(x)
            except:
                raise TypeError("Audio cannot be casted to tensor. Cast it manually")

        window_size_samples = len(x[0]) if x.dim() == 2 else len(x)
        self.current_sample += window_size_samples

        speech_prob = self.model(x, self.sampling_rate).item()

        if (speech_prob >= self.threshold) and self.temp_end:
            self.temp_end = 0

        if (speech_prob >= self.threshold) and not self.triggered:
            self.triggered = True
            speech_start = self.current_sample - self.speech_pad_samples
            return {
                "start": (
                    int(speech_start)
                    if not return_seconds
                    else round(speech_start / self.sampling_rate, 1)
                )
            }

        if (speech_prob < self.threshold - 0.15) and self.triggered:
            if not self.temp_end:
                self.temp_end = self.current_sample
            if self.current_sample - self.temp_end < self.min_silence_samples:
                return None
            else:
                speech_end = self.temp_end + self.speech_pad_samples
                self.temp_end = 0
                self.triggered = False
                return {
                    "end": (
                        int(speech_end)
                        if not return_seconds
                        else round(speech_end / self.sampling_rate, 1)
                    )
                }

        return None


#######################
# because Silero now requires exactly 512-sized audio chunks

import numpy as np


class FixedVADIterator(VADIterator):
    """It fixes VADIterator by allowing to process any audio length, not only exactly 512 frames at once.
    If audio to be processed at once is long and multiple voiced segments detected,
    then __call__ returns the start of the first segment, and end (or middle, which means no end) of the last segment.
    """

    def reset_states(self):
        super().reset_states()
        self.buffer = np.array([], dtype=np.float32)

    def __call__(self, x, return_seconds=False):
        self.buffer = np.append(self.buffer, x)
        ret = None
        while len(self.buffer) >= 512:
            r = super().__call__(self.buffer[:512], return_seconds=return_seconds)
            self.buffer = self.buffer[512:]
            if ret is None:
                ret = r
            elif r is not None:
                if "end" in r:
                    ret["end"] = r["end"]  # the latter end
                if "start" in r and "end" in ret:  # there is an earlier start.
                    # Remove end, merging this segment with the previous one.
                    del ret["end"]
        return ret if ret != {} else None


if __name__ == "__main__":
    # test/demonstrate the need for FixedVADIterator:

    import torch

    model, _ = torch.hub.load(repo_or_dir="snakers4/silero-vad", model="silero_vad")
    vac = FixedVADIterator(model)
    #   vac = VADIterator(model)  # the second case crashes with this

    # this works: for both
    audio_buffer = np.array([0] * (512), dtype=np.float32)
    vac(audio_buffer)

    # this crashes on the non FixedVADIterator with
    # ops.prim.RaiseException("Input audio chunk is too short", "builtins.ValueError")
    audio_buffer = np.array([0] * (512 - 1), dtype=np.float32)
    vac(audio_buffer)