ten-vad-model-config.h
1.1 KB
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
// sherpa-onnx/csrc/ten-vad-model-config.h
//
// Copyright (c) 2025 Xiaomi Corporation
#ifndef SHERPA_ONNX_CSRC_TEN_VAD_MODEL_CONFIG_H_
#define SHERPA_ONNX_CSRC_TEN_VAD_MODEL_CONFIG_H_
#include <string>
#include "sherpa-onnx/csrc/parse-options.h"
namespace sherpa_onnx {
struct TenVadModelConfig {
std::string model;
// threshold to classify a segment as speech
//
// If the predicted probability of a segment is larger than this
// value, then it is classified as speech.
float threshold = 0.5;
float min_silence_duration = 0.5; // in seconds
float min_speech_duration = 0.25; // in seconds
// 160 or 256
int32_t window_size = 256; // in samples
// If a speech segment is longer than this value, then we increase
// the threshold to 0.9. After finishing detecting the segment,
// the threshold value is reset to its original value.
float max_speech_duration = 20; // in seconds
TenVadModelConfig() = default;
void Register(ParseOptions *po);
bool Validate() const;
std::string ToString() const;
};
} // namespace sherpa_onnx
#endif // SHERPA_ONNX_CSRC_TEN_VAD_MODEL_CONFIG_H_