Files
OpsLog/internal/audio/manager.go
T
rouggy 4fe5811fd2 fix: a playback that never starts no longer fails silently
From an operator's log, playing a recording on the air repeatedly:

  18:52:40.148 play → 18:52:48.420 PTT released   8.2 s, the whole take
  18:52:57.759 play → 18:52:57.879 PTT released   0.12 s, nothing came out

Those 120 ms are the PTT release tail alone: the render device refused to start
and playPCM returned an error that was thrown away. The call reported success
every time, which is what "it plays once and then never again" looks like from
the outside — and why the search went to the recorder, which was innocent.

The error is now logged with the device name.
2026-07-31 18:55:48 +02:00

303 lines
8.8 KiB
Go

//go:build windows
package audio
import (
"fmt"
"sync"
)
// Manager owns the DVK record/playback lifecycle: at most one recording and
// one playback at a time. Device ids are passed per call so the host can route
// recording to the mic and playback to the rig (or the preview speakers).
type Manager struct {
mu sync.Mutex
recStop chan struct{}
recDone chan recResult
playStop chan struct{}
monStop chan struct{} // RX monitor passthrough (capture → render)
monRing *pcmRing // live audio hand-off, also fed by the network stream
txStop chan struct{} // TX audio passthrough (mic → rig)
onChange func() // fired on any record/playback state transition
}
type recResult struct {
pcm []byte
err error
}
// NewManager creates a DVK manager. onChange (optional) is called whenever the
// recording/playback state changes, so the host can push an audio:status event.
func NewManager(onChange func()) *Manager { return &Manager{onChange: onChange} }
func (m *Manager) notify() {
if m.onChange != nil {
m.onChange()
}
}
// StartRecording begins capturing from deviceID into memory. Finish with
// StopRecording (which writes the WAV) or CancelRecording (which discards it).
func (m *Manager) StartRecording(deviceID string) error {
m.mu.Lock()
if m.recStop != nil {
m.mu.Unlock()
return fmt.Errorf("already recording")
}
stop := make(chan struct{})
done := make(chan recResult, 1)
m.recStop, m.recDone = stop, done
m.mu.Unlock() // release BEFORE notify — onChange re-enters via IsRecording()
go func() {
pcm, err := recordPCM(deviceID, stop)
done <- recResult{pcm, err}
}()
m.notify()
return nil
}
// StopRecording ends the capture and writes it to path as a WAV file.
func (m *Manager) StopRecording(path string) error {
m.mu.Lock()
stop, done := m.recStop, m.recDone
m.recStop, m.recDone = nil, nil
m.mu.Unlock()
if stop == nil {
return fmt.Errorf("not recording")
}
close(stop)
res := <-done
m.notify()
if res.err != nil {
return res.err
}
if len(res.pcm) == 0 {
return fmt.Errorf("captured no audio (check the recording device)")
}
return writeWAV(path, res.pcm)
}
// CancelRecording aborts a recording without saving.
func (m *Manager) CancelRecording() {
m.mu.Lock()
stop, done := m.recStop, m.recDone
m.recStop, m.recDone = nil, nil
m.mu.Unlock()
if stop != nil {
close(stop)
<-done
m.notify()
}
}
func (m *Manager) IsRecording() bool {
m.mu.Lock()
defer m.mu.Unlock()
return m.recStop != nil
}
func (m *Manager) IsPlaying() bool {
m.mu.Lock()
defer m.mu.Unlock()
return m.playStop != nil
}
// Play renders a WAV file to deviceID. Any current playback is stopped first.
// Returns immediately; playback runs in the background.
// Play sends a recorded message to a device, amplified by gainPct (100 = as
// recorded).
//
// The gain exists because nothing else could raise the level: the message went
// out exactly as captured, so a mic recorded quietly drove the rig quietly and
// the operator had no control anywhere in OpsLog — only the radio's own USB
// input level, buried in its menus, and the Windows mixer.
func (m *Manager) Play(deviceID, path string, gainPct int) error {
pcm, rate, ch, bits, err := readWAV(path)
if err != nil {
return err
}
if gainPct > 0 && gainPct != 100 && bits == 16 {
g := float64(gainPct) / 100
// In place: the buffer is this call's own copy of the file.
for i := 0; i+1 < len(pcm); i += 2 {
v := int16(uint16(pcm[i]) | uint16(pcm[i+1])<<8)
v = scalePCM(v, g)
pcm[i], pcm[i+1] = byte(uint16(v)), byte(uint16(v)>>8)
}
}
m.StopPlayback()
stop := make(chan struct{})
m.mu.Lock()
m.playStop = stop
m.mu.Unlock()
go func() {
// The error was discarded. A device that refuses to start returns here
// instantly, the PTT is released 120 ms later, and NOTHING says why —
// which is exactly what a station heard as "it plays once, then never
// again": the call succeeded, the sound did not.
if err := playPCM(deviceID, pcm, rate, ch, bits, stop); err != nil {
LogSink("audio: playback on %q failed: %v", DeviceName(deviceID), err)
}
m.mu.Lock()
if m.playStop == stop {
m.playStop = nil
}
m.mu.Unlock()
m.notify()
}()
m.notify()
return nil
}
// StopPlayback halts any in-progress playback.
func (m *Manager) StopPlayback() {
m.mu.Lock()
stop := m.playStop
m.playStop = nil
m.mu.Unlock()
if stop != nil {
close(stop)
m.notify()
}
}
// ---- RX audio monitor (Phase 2: USB codec passthrough) --------------------
//
// StartMonitor pipes live RX audio from inputDev (e.g. the rig's "USB Audio
// CODEC" capture endpoint) to outputDev (your speakers/headset) through a
// latency-bounded ring, so you HEAR the radio inside OpsLog. The very same ring
// is later fed by the network 50003 stream instead of a USB capture — the render
// half is transport-agnostic. inputDev "" = system default capture.
func (m *Manager) StartMonitor(inputDev, outputDev string) error {
return m.startMonitor(inputDev, outputDev, true)
}
// StartMonitorSink starts ONLY the render side (no USB capture) so an external
// producer — the network 50003 stream — can feed decoded RX PCM via
// PushMonitorAudio. Same output path as StartMonitor, minus the capture goroutine.
func (m *Manager) StartMonitorSink(outputDev string) error {
return m.startMonitor("", outputDev, false)
}
// startMonitor wires the RX monitor: always a render loop pulling from monRing;
// when capture is true it also captures inputDev into that ring (USB monitor).
// When false the ring is fed only by PushMonitorAudio (network audio).
func (m *Manager) startMonitor(inputDev, outputDev string, capture bool) error {
m.mu.Lock()
if m.monStop != nil {
m.mu.Unlock()
return fmt.Errorf("monitor already running")
}
stop := make(chan struct{})
ring := newPCMRing(bytesPerSec / 2) // ~500 ms cap — low latency for live monitor
m.monStop, m.monRing = stop, ring
m.mu.Unlock()
if capture {
// Producer: capture the rig's USB audio into the ring.
go func() {
_ = captureStream(inputDev, stop, func(chunk []byte) { ring.Push(chunk) })
}()
}
// Consumer: render the ring to the output device at the internal 16 kHz mono.
go func() {
_ = renderStream(outputDev, sampleRate, channels, bitsPerSample, stop, ring)
}()
m.notify()
return nil
}
// StopMonitor stops the RX monitor passthrough.
func (m *Manager) StopMonitor() {
m.mu.Lock()
stop := m.monStop
m.monStop, m.monRing = nil, nil
m.mu.Unlock()
if stop != nil {
close(stop)
m.notify()
}
}
// MonitorActive reports whether the RX monitor passthrough is running.
func (m *Manager) MonitorActive() bool {
m.mu.Lock()
defer m.mu.Unlock()
return m.monStop != nil
}
// PushMonitorAudio feeds externally-sourced PCM (16 kHz mono 16-bit) into the
// active monitor's output — the hook the network 50003 audio stream uses to play
// decoded RX through the very same output path a USB capture feeds. No-op when no
// monitor is running. Keeps the unexported ring inside the package.
func (m *Manager) PushMonitorAudio(pcm []byte) {
m.mu.Lock()
ring := m.monRing
m.mu.Unlock()
if ring != nil {
ring.Push(pcm)
}
}
// ---- TX audio passthrough (Phase 3: live mic → rig over USB) --------------
//
// StartTXAudio pipes your live microphone (micDev) into the rig's audio input
// (toRadioDev — for a USB-connected rig, its "USB Audio CODEC" render endpoint),
// so you talk through the PC. It is the mirror of StartMonitor (same ring +
// capture + render primitives, source/sink swapped). PTT keying is the caller's
// job (the app layer keys PTT before this and unkeys after) so this stays a pure
// audio route. The captured 16 kHz mono stream is also the exact shape the future
// network 50003 TX will encode and send — so Phase 5 reuses this capture side.
func (m *Manager) StartTXAudio(micDev, toRadioDev string) error {
m.mu.Lock()
if m.txStop != nil {
m.mu.Unlock()
return fmt.Errorf("TX audio already running")
}
stop := make(chan struct{})
ring := newPCMRing(bytesPerSec / 4) // ~250 ms — tighter for live TX latency
m.txStop = stop
m.mu.Unlock()
go func() {
_ = captureStream(micDev, stop, func(chunk []byte) { ring.Push(chunk) })
}()
go func() {
_ = renderStream(toRadioDev, sampleRate, channels, bitsPerSample, stop, ring)
}()
m.notify()
return nil
}
// StopTXAudio stops the TX mic→rig passthrough.
func (m *Manager) StopTXAudio() {
m.mu.Lock()
stop := m.txStop
m.txStop = nil
m.mu.Unlock()
if stop != nil {
close(stop)
m.notify()
}
}
// TXAudioActive reports whether the TX mic→rig passthrough is running.
func (m *Manager) TXAudioActive() bool {
m.mu.Lock()
defer m.mu.Unlock()
return m.txStop != nil
}
// scalePCM applies a gain to one sample, clamping rather than wrapping — an
// overflow that wraps turns loud speech into a burst of noise on the air.
func scalePCM(s int16, g float64) int16 {
v := float64(s) * g
if v > 32767 {
return 32767
}
if v < -32768 {
return -32768
}
return int16(v)
}