speakrs 0.6.0

Fast Rust speaker diarization with pyannote-level accuracy and native CPU/CoreML/CUDA inference
Documentation 
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

use ndarray::Array2;
use tracing::debug;

use crate::binarize::ActivityCleanup;
use crate::clustering::plda::PldaTransform;
use crate::reconstruct::{ExclusiveDiarization, Reconstructor};
use crate::segment::{merge_exclusive_segments, merge_segments};

use super::config::{PipelineConfig, ReconstructMethod};
use super::types::{
    ChunkSpeakerClusters, DiarizationResult, DiscreteDiarization, InferenceArtifacts, PipelineError,
};

/// Run clustering and reconstruction on pre-computed inference artifacts
pub fn post_inference(
    inference_artifacts: InferenceArtifacts,
    config: &PipelineConfig,
    plda: &PldaTransform,
) -> Result<DiarizationResult, PipelineError> {
    let post_start = std::time::Instant::now();
    let InferenceArtifacts {
        layout,
        segmentations,
        embeddings,
        #[cfg(feature = "_metrics")]
            stage_timings: _,
    } = inference_artifacts;
    let speaker_count = segmentations.speaker_count(&layout);

    if speaker_count
        .iter()
        .all(|speaker_count| *speaker_count == 0)
    {
        return Ok(DiarizationResult {
            segmentations,
            embeddings,
            speaker_count,
            hard_clusters: ChunkSpeakerClusters(Array2::zeros((0, 0))),
            discrete_diarization: DiscreteDiarization(Array2::zeros((0, 0))),
            segments: Vec::new(),
            exclusive_segments: Vec::new(),
        });
    }

    let training_embeddings =
        embeddings.training_set(&segmentations, config.effective_clean_frame_duration());
    let hard_clusters = training_embeddings.cluster(&segmentations, &embeddings, plda, config)?;

    let reconstructor = Reconstructor::new(&segmentations, &hard_clusters, &layout.start_frames)?;
    let activations = reconstructor.frame_activations(&speaker_count);
    let discrete_diarization = match config.reconstruct_method {
        ReconstructMethod::Smoothed { epsilon } => {
            reconstructor.reconstruct_smoothed_with(&activations, &speaker_count, epsilon)
        }
        ReconstructMethod::Standard => reconstructor.reconstruct_with(&activations, &speaker_count),
    };

    let discrete_diarization = apply_activity_cleanup(discrete_diarization, config.activity);
    let exclusive_diarization =
        ExclusiveDiarization::from_scored(&discrete_diarization, &activations)?;

    let segments = discrete_diarization.to_segments();
    let segments = merge_segments(&segments, config.merge_gap);
    let exclusive_segments = exclusive_diarization.to_segments();
    let exclusive_segments = merge_exclusive_segments(&exclusive_segments, config.merge_gap);

    debug!(
        post_inference_ms = post_start.elapsed().as_millis(),
        "Post-inference complete"
    );

    Ok(DiarizationResult {
        segmentations,
        embeddings,
        speaker_count,
        hard_clusters,
        discrete_diarization,
        segments,
        exclusive_segments,
    })
}

pub(super) fn apply_activity_cleanup(
    discrete: DiscreteDiarization,
    config: ActivityCleanup,
) -> DiscreteDiarization {
    if config.is_identity() {
        discrete
    } else {
        DiscreteDiarization(config.apply(&discrete))
    }
}

#[cfg(test)]
mod tests {
    use ndarray::array;

    use super::apply_activity_cleanup;
    use crate::binarize::ActivityCleanup;
    use crate::pipeline::{DiscreteDiarization, FrameActivations};
    use crate::reconstruct::ExclusiveDiarization;
    use crate::reconstruct::test_support::exclusive_as_discrete;

    #[test]
    fn default_cleanup_is_identity() {
        assert!(ActivityCleanup::default().is_identity());
        let discrete = DiscreteDiarization(array![[0.0], [1.0], [0.0]]);
        let cleaned = apply_activity_cleanup(discrete.clone(), ActivityCleanup::default());
        assert_eq!(&*cleaned, &*discrete);
    }

    #[test]
    fn padding_only_cleanup_is_applied() {
        let config = ActivityCleanup::new(0, 0, 1, 1);
        assert!(!config.is_identity());
        let discrete = DiscreteDiarization(array![[0.0], [0.0], [1.0], [0.0], [0.0]]);
        let cleaned = apply_activity_cleanup(discrete, config);
        assert_eq!(&*cleaned, &array![[0.0], [1.0], [1.0], [1.0], [0.0]]);
    }

    #[test]
    fn cleanup_before_exclusive_selection_preserves_surviving_speech() {
        let full = DiscreteDiarization(array![
            [0.0, 0.0],
            [0.0, 1.0],
            [1.0, 1.0],
            [0.0, 1.0],
            [0.0, 0.0],
        ]);
        let activations = FrameActivations(array![
            [0.0, 0.0],
            [0.1, 0.8],
            [0.9, 0.2],
            [0.1, 0.8],
            [0.0, 0.0],
        ]);
        let cleaned = apply_activity_cleanup(full, ActivityCleanup::new(2, 0, 0, 0));

        let exclusive = ExclusiveDiarization::from_scored(&cleaned, &activations).unwrap();
        assert_eq!(
            &exclusive_as_discrete(&exclusive).0,
            &array![[0.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 0.0],]
        );

        for (cleaned_row, exclusive_row) in cleaned
            .rows()
            .into_iter()
            .zip(exclusive_as_discrete(&exclusive).rows())
        {
            assert_eq!(
                cleaned_row.iter().any(|value| *value > 0.0),
                exclusive_row.iter().any(|value| *value > 0.0)
            );
        }
    }
}