more vad tests

Author: yujonglee

Cargo.lock

@@ -43,6 +43,7 @@ hypr-stt = { path = "crates/stt", package = "stt", features = ["realtime", "reco
 hypr-template = { path = "crates/template", package = "template" }
 hypr-timeline = { path = "crates/timeline", package = "timeline" }
 hypr-turso = { path = "crates/turso", package = "turso" }
+hypr-vad = { path = "crates/vad", package = "vad" }
 hypr-whisper = { path = "crates/whisper", package = "whisper" }
 hypr-ws = { path = "crates/ws", package = "ws" }
 hypr-ws-utils = { path = "crates/ws-utils", package = "ws-utils" }

Cargo.toml

@@ -8,6 +8,7 @@ hound = { workspace = true }
 hypr-data = { workspace = true }
 
 [dependencies]
+hypr-vad = { workspace = true }
 kalosm-sound = { workspace = true, default-features = false }
 rodio = { workspace = true, features = ["wav"] }
 

crates/chunker/Cargo.toml

@@ -10,3 +10,6 @@ tracing = { workspace = true }
 
 ndarray = "0.16"
 ort = "=2.0.0-rc.9"
+
+[dev-dependencies]
+hypr-data = { workspace = true }

crates/vad/Cargo.toml

@@ -110,10 +110,27 @@ mod tests {
     use super::*;
 
     #[test]
-    fn test_vad() {
+    fn test_vad_silence() {
         let mut vad = Vad::new().unwrap();
         let audio_samples = vec![0.0; 16000];
         let prob = vad.run(&audio_samples).unwrap();
         assert!(prob < 0.1);
     }
+
+    #[test]
+    fn test_vad_long() {
+        let mut vad = Vad::new().unwrap();
+        let audio_samples = to_f32(hypr_data::english_2::AUDIO);
+        let prob = vad.run(&audio_samples).unwrap();
+        assert!(prob > 0.8);
+    }
+
+    fn to_f32(bytes: &[u8]) -> Vec<f32> {
+        let mut samples = Vec::with_capacity(bytes.len() / 2);
+        for chunk in bytes.chunks_exact(2) {
+            let sample = i16::from_le_bytes([chunk[0], chunk[1]]) as f32 / 32768.0;
+            samples.push(sample);
+        }
+        samples
+    }
 }