feat(audio): add write_f32() and migrate PlaybackEngine to f32 pipeline

PlaybackEngine::append() now accepts Source<Item = f32>.
ALSA Direct thread uses f32 buffer and calls write_f32().

New write_f32() method in AlsaDirectStream converts f32 samples to
hardware format with full 24-bit precision:
- S24_3LE: f32 -> 24-bit packed 3 bytes (SMSL/Topping DACs)
- S32LE: f32 -> i32 full range
- S24LE: f32 -> 24-bit in i32 container
- S16LE: f32 -> i16
- FloatLE: direct write

Old write(&[i16]) kept for safety during migration.
This commit is contained in:
vicrodh committed 2026-02-05 16:19:27 -06:00
1 parent 2c276154b8
commit a923798bae
2 files changed
+172 -13

No files matched your search

+11 -12
View File
@@ -56,7 +56,7 @@ impl PlaybackEngine {
/// Append audio source
pub fn append<S>(&mut self, source: S) -> Result<(), String>
where
S: Source<Item = i16> + Send + 'static,
S: Source<Item = f32> + Send + 'static,
{
match self {
Self::Rodio { sink } => {
@@ -73,11 +73,10 @@ impl PlaybackEngine {
hardware_volume: _,
} => {
// For ALSA Direct, we need to spawn a thread that:
// 1. Streams samples from source (no buffering entire file)
// 2. Converts i16 samples to f32
// 3. Writes to ALSA PCM
// 4. Tracks position
// 5. Supports pause/resume without terminating
// 1. Streams f32 samples from source (no buffering entire file)
// 2. Writes to ALSA PCM (converts f32 to hardware format)
// 3. Tracks position
// 4. Supports pause/resume without terminating
let stream_clone = stream.clone();
let is_playing_clone = is_playing.clone();
@@ -98,7 +97,7 @@ impl PlaybackEngine {
const CHUNK_SIZE: usize = 8192; // frames per chunk
let chunk_samples = CHUNK_SIZE * channels as usize;
let mut buffer_i16 = Vec::with_capacity(chunk_samples);
let mut buffer_f32 = Vec::with_capacity(chunk_samples);
let mut total_frames: u64 = 0;
let mut source_iter = source.into_iter();
@@ -123,15 +122,15 @@ impl PlaybackEngine {
}
// Fill buffer from source
buffer_i16.clear();
buffer_f32.clear();
for _ in 0..chunk_samples {
match source_iter.next() {
Some(sample) => buffer_i16.push(sample),
Some(sample) => buffer_f32.push(sample),
None => break, // End of stream
}
}
if buffer_i16.is_empty() {
if buffer_f32.is_empty() {
// End of stream
log::info!("[ALSA Direct Engine] Stream ended (total frames: {})", total_frames);
natural_end = true;
@@ -140,13 +139,13 @@ impl PlaybackEngine {
// Write to ALSA (auto-converts based on detected format)
// This is bit-perfect: no resampling, no mixing, direct to hardware
if let Err(e) = stream_clone.write(&buffer_i16) {
if let Err(e) = stream_clone.write_f32(&buffer_f32) {
log::error!("[ALSA Direct Engine] Write failed: {}", e);
break 'playback;
}
// Update position
let frames_written = buffer_i16.len() / channels as usize;
let frames_written = buffer_f32.len() / channels as usize;
total_frames += frames_written as u64;
position_clone.store(total_frames, Ordering::SeqCst);
duration_clone.store(total_frames, Ordering::SeqCst);