features.h
1.9 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
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
// sherpa-onnx/csrc/features.h
//
// Copyright (c) 2023 Xiaomi Corporation
#ifndef SHERPA_ONNX_CSRC_FEATURES_H_
#define SHERPA_ONNX_CSRC_FEATURES_H_
#include <memory>
#include <string>
#include <vector>
namespace sherpa_onnx {
struct FeatureExtractorConfig {
float sampling_rate = 16000;
int32_t feature_dim = 80;
std::string ToString() const;
};
class FeatureExtractor {
public:
explicit FeatureExtractor(const FeatureExtractorConfig &config = {});
~FeatureExtractor();
/**
@param sampling_rate The sampling_rate of the input waveform. Should match
the one expected by the feature extractor.
@param waveform Pointer to a 1-D array of size n
@param n Number of entries in waveform
*/
void AcceptWaveform(float sampling_rate, const float *waveform, int32_t n);
/**
* InputFinished() tells the class you won't be providing any
* more waveform. This will help flush out the last frame or two
* of features, in the case where snip-edges == false; it also
* affects the return value of IsLastFrame().
*/
void InputFinished();
int32_t NumFramesReady() const;
/** Note: IsLastFrame() will only ever return true if you have called
* InputFinished() (and this frame is the last frame).
*/
bool IsLastFrame(int32_t frame) const;
/** Get n frames starting from the given frame index.
*
* @param frame_index The starting frame index
* @param n Number of frames to get.
* @return Return a 2-D tensor of shape (n, feature_dim).
* which is flattened into a 1-D vector (flattened in in row major)
*/
std::vector<float> GetFrames(int32_t frame_index, int32_t n) const;
void Reset();
/// Return feature dim of this extractor
int32_t FeatureDim() const;
private:
class Impl;
std::unique_ptr<Impl> impl_;
};
} // namespace sherpa_onnx
#endif // SHERPA_ONNX_CSRC_FEATURES_H_