rvsttd: hold-back last drain chunk for fade, same faded audio to Moonshine + WAV

This commit is contained in:
2026-08-17 06:58:18 +00:00
parent cc9fc95519
commit b8ee014cc3
+28 -31
View File
@@ -113,14 +113,6 @@ impl DebugRecorder {
self.audio.extend_from_slice(samples);
}
fn audio_len(&self) -> usize {
self.audio.len()
}
fn truncate_audio(&mut self, len: usize) {
self.audio.truncate(len);
}
fn save(self) {
// Write log
let log_path = self.dir.join("session.log");
@@ -332,45 +324,50 @@ fn transcriber_loop(
// Drain trailing audio from ALSA buffer (cpal stream still alive).
// Fixed 100ms window — cpal delivers every ~50ms (800 samples @ 16kHz),
// so this captures 1-2 more callbacks worth of trailing audio.
// Hold one chunk back so we can apply a fade-out to the very last one.
let drain_deadline = std::time::Instant::now() + Duration::from_millis(100);
let mut drained: Vec<f32> = Vec::new();
let mut held_chunk: Option<Vec<f32>> = None;
while std::time::Instant::now() < drain_deadline {
match rx.recv_timeout(drain_deadline - std::time::Instant::now()) {
Ok(chunk) => {
if let Some(ref mut r) = recorder {
r.add_audio(&chunk);
// Feed previously held chunk to Moonshine + recorder (no fade)
if let Some(prev) = held_chunk.take() {
if let Some(ref mut r) = recorder {
r.add_audio(&prev);
}
unsafe {
moonshine_transcribe_add_audio_to_stream(
handle, stream_handle,
prev.as_ptr(), prev.len() as u64,
SAMPLE_RATE, 0,
);
}
}
drained.extend_from_slice(&chunk);
held_chunk = Some(chunk);
}
Err(mpsc::RecvTimeoutError::Timeout) => break,
Err(mpsc::RecvTimeoutError::Disconnected) => break,
}
}
// Apply 5ms fade-out to the trailing edge of the drained audio.
// Both the debug WAV and Moonshine receive the same faded audio.
let fade_samples = (SAMPLE_RATE as usize * 5) / 1000;
if drained.len() > fade_samples {
let start = drained.len() - fade_samples;
for i in 0..fade_samples {
let t = 1.0 - (i as f32 / fade_samples as f32);
drained[start + i] *= t;
// Apply 5ms fade-out to the last chunk, then feed to both Moonshine and recorder
if let Some(mut chunk) = held_chunk.take() {
let fade_samples = (SAMPLE_RATE as usize * 5) / 1000;
if chunk.len() > fade_samples {
let start = chunk.len() - fade_samples;
for i in 0..fade_samples {
let t = 1.0 - (i as f32 / fade_samples as f32);
chunk[start + i] *= t;
}
}
}
// Update recorder with the faded audio (replaces the raw versions added above)
if let Some(ref mut r) = recorder {
// Remove the raw drained samples and re-add the faded ones
let raw_drained_len = r.audio_len();
r.truncate_audio(raw_drained_len.saturating_sub(drained.len()));
r.add_audio(&drained);
}
if !drained.is_empty() {
if let Some(ref mut r) = recorder {
r.add_audio(&chunk);
}
unsafe {
moonshine_transcribe_add_audio_to_stream(
handle, stream_handle,
drained.as_ptr(), drained.len() as u64,
chunk.as_ptr(), chunk.len() as u64,
SAMPLE_RATE, 0,
);
}