polyvoice 0.6.2

Speaker diarization library for Rust — online and offline, ONNX-powered, ecosystem-agnostic
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

---
schema_version: 1
kind: module_contract
module: src/vad
level: subsystem
layer: algorithm
purpose: >
  Owns the VoiceActivityDetector trait, energy-based VAD, VAD state machine,
  and speech segmentation utilities. Does NOT own ONNX-based SileroVAD
  (that lives in silero_vad.rs).
status: stable
owners:
  - polyvoice-core
workcell:
  type: leaf
  parent: ""
  children: []
  owns_paths:
    - src/vad/
  context_budget:
    max_files: 12
    max_source_lines: 1500
    max_contract_lines: 180
    max_readme_lines: 120
    max_todo_lines: 80
authority:
  write_policy: single_active_write_lease
  orchestrator: polyvoice-core
  read_agents: many_allowed
  migration_lease_required:
    - cross-workcell write
    - public surface migration
surface:
  - name: VoiceActivityDetector
    kind: trait
    visibility: public
    contract: >
      Core VAD trait: process audio samples → speech/non-speech decisions.
    proof:
      kind: unit-test
      target: src/vad::mod::tests
      command: cargo test --lib vad
  - name: EnergyVad
    kind: struct
    visibility: public
    contract: >
      Simple energy-threshold VAD implementation.
    proof:
      kind: unit-test
      target: src/vad::mod::tests
      command: cargo test --lib vad
  - name: VadConfig
    kind: struct
    visibility: public
    contract: >
      Configuration for VAD parameters (thresholds, frame sizes).
    proof:
      kind: unit-test
      target: src/vad::mod::tests
      command: cargo test --lib vad
  - name: segment_speech
    kind: function
    visibility: public
    contract: >
      Segments audio into speech regions using a VAD driver.
    proof:
      kind: unit-test
      target: src/vad::mod::tests
      command: cargo test --lib vad
  - name: VadStateMachine
    kind: struct
    visibility: public
    contract: >
      Hysteresis state machine for VAD decision smoothing.
    proof:
      kind: unit-test
      target: src/vad::mod::tests
      command: cargo test --lib vad
dependencies:
  internal:
    - module: types
      scope: data-shape
      reason: DiarizationConfig for VAD parameters.
  external: []
consumers:
  - path: src/pipeline/mod.rs
    uses:
      - VoiceActivityDetector
      - VadConfig
      - segment_speech
  - path: src/streaming/mod.rs
    uses:
      - VoiceActivityDetector
      - VadConfig
      - VadEvent
      - VadStateMachine
  - path: src/silero_vad/mod.rs
    uses:
      - VoiceActivityDetector
  - path: src/ffi/mod.rs
    uses:
      - VadConfig
  - path: tests/vad_test.rs
    uses:
      - EnergyVad
invariants:
  - id: vad-monotonic
    rule: segment_speech returns non-overlapping, monotonically ordered segments.
    proof:
      kind: unit-test
      target: src/vad::mod::tests
      command: cargo test --lib vad
verification:
  pre_change:
    - cargo test --lib vad
  full:
    - cargo test --lib vad
    - cargo test --test vad_test
    - cargo clippy --all-targets --all-features -- -D warnings
agent_policy:
  allowed_mutations:
    - Tuning energy threshold heuristics.
    - Adding new VAD implementations.
  forbidden_mutations:
    - Changing VoiceActivityDetector trait without migration lease.
  escalation:
    - Changes to VoiceActivityDetector trait or associated types.
---

# src/vad

Voice Activity Detection trait, energy-based VAD, and speech segmentation.