lwh2017 commited on
Commit
9f3dbd6
·
verified ·
1 Parent(s): fac3b17

Add README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -1
README.md CHANGED
@@ -35,9 +35,16 @@ IB-Robot computes these on the host in
35
  `src/voice_asr_service/voice_asr_service/speaker_id/features.py`, which is pinned
36
  against sherpa-onnx's own extractor by `test_speaker_features.py`.
37
 
 
 
 
 
 
 
 
38
  | Tensor | Semantic | Shape | dtype |
39
  |---|---|---|---|
40
- | input | `host.campplus.features` | `(1, T, 80)` | float32 |
41
  | output | `host.campplus.embedding` | `(1, 192)` | float32 |
42
 
43
  ## Deployments
 
35
  `src/voice_asr_service/voice_asr_service/speaker_id/features.py`, which is pinned
36
  against sherpa-onnx's own extractor by `test_speaker_features.py`.
37
 
38
+ The host also splits an utterance into **fixed 200-frame (2.0 s) windows**, which
39
+ is what the input shape above reflects. Shorter utterances are tiled from their
40
+ own frames rather than zero-padded; longer ones become consecutive windows with
41
+ the last aligned to the end, and the per-window unit vectors are averaged. This
42
+ is part of the declared preprocessing contract, so every deployment applies it
43
+ and vectors stay comparable across them.
44
+
45
  | Tensor | Semantic | Shape | dtype |
46
  |---|---|---|---|
47
+ | input | `host.campplus.features` | `(1, 200, 80)` | float32 |
48
  | output | `host.campplus.embedding` | `(1, 192)` | float32 |
49
 
50
  ## Deployments