trying a new method

pi/bin/video-publisher.sh

@@ -20,27 +20,11 @@ VIDEO_FPS="${VIDEO_FPS:-30}"
 VIDEO_BITRATE="${VIDEO_BITRATE:-3000000}"
 AUDIO_ENABLE="${AUDIO_ENABLE:-0}"
 AUDIO_DEVICE="${AUDIO_DEVICE:-plughw:0,0}"
-AUDIO_RATE="${AUDIO_RATE:-48000}"
+AUDIO_CODEC="pcm_s16le"
-AUDIO_CHANNELS="${AUDIO_CHANNELS:-1}"
+# Keep the audio path simple for the Pi Zero; we hardcode a single PCM profile.
-AUDIO_BITRATE="${AUDIO_BITRATE:-24000}"
+AUDIO_RATE=16000
-AUDIO_CODEC="${AUDIO_CODEC:-pcm_mulaw}"
-
-# Normalize device/rate to match the HAT capture; plughw handles any minor conversions.
-AUDIO_DEVICE="plughw:0,0"
 AUDIO_CHANNELS=1
-# Defaults chosen to keep Pi Zero CPU low. Opus can be enabled by setting AUDIO_CODEC=libopus.
+AUDIO_DEVICE="plughw:0,0"
-if [[ "${AUDIO_CODEC}" == "libopus" ]]; then
-	AUDIO_RATE=16000
-	AUDIO_BITRATE=16000
-else
-	# G.711 mu-law (or A-law if AUDIO_CODEC=pcm_alaw): very low CPU, 8 kHz mono, ~64 kbps.
-	AUDIO_CODEC="pcm_mulaw"
-	AUDIO_RATE=8000
-	AUDIO_BITRATE=64000
-fi
-
-# Disable audio streaming for now; only video will be published.
-AUDIO_ENABLE=0
 # Flip the camera 180deg (supported by rpicam-vid/libcamera-vid)
 FLIP_ARGS=(--rotation 180)
 
@@ -64,9 +48,6 @@ else
 	exit 1
 fi
 
-# Flip using the native rotation flag (0/180 supported per rpicam-vid --help).
-FLIP_ARGS=(--rotation 180)
-
 run_pipeline() {
 	if [[ "${AUDIO_ENABLE}" -eq 1 ]]; then
 		# Try audio + video; if audio device is missing, fallback to video-only.
@@ -101,11 +82,9 @@ run_pipeline() {
 				-i "${AUDIO_DEVICE}" \
 				-c:v copy \
 				-c:a "${AUDIO_CODEC}" \
-				-b:a "${AUDIO_BITRATE}" \
-				$( [[ "${AUDIO_CODEC}" == "libopus" ]] && echo "-compression_level 0 -application voip -frame_duration 60" ) \
-				-ac:a "${AUDIO_CHANNELS}" \
 				-ar:a "${AUDIO_RATE}" \
-				-af "$( [[ "${AUDIO_CODEC}" == "libopus" ]] && echo "pan=1c|c0=c0," )volume=20dB" \
+				-ac:a "${AUDIO_CHANNELS}" \
+				-af "volume=20dB" \
 				-flush_packets 1 \
 				-f mpegts \
 				"${PUBLISH_URL}"

pi/install_roverd.sh

@@ -204,8 +204,10 @@ VIDEO_WIDTH=1280
 VIDEO_HEIGHT=720
 VIDEO_FPS=30
 VIDEO_BITRATE=2000000
-AUDIO_ENABLE=1
 AUDIO_ENABLE=0
+AUDIO_DEVICE=plughw:0,0
+AUDIO_RATE=16000
+AUDIO_CHANNELS=1
 ENV
 chown roverd:roverd /var/lib/roverd/video.env
 chmod 0640 /var/lib/roverd/video.env

pi/roverd/config.go

@@ -143,7 +143,7 @@ func LoadConfig(path string) (*Config, error) {
 		Audio: AudioConfig{
 			CaptureEnabled: false,
 			CaptureDevice:  "plughw:0,0",
-			SampleRate:     48000,
+			SampleRate:     16000,
 			Channels:       1,
 			Bitrate:        24000,
 			TTSEnabled:     false,

pi/roverd/media_env.go

@@ -25,8 +25,20 @@ func UpdatePublisherEnv(media MediaConfig, audio AudioConfig) error {
 	fmt.Fprintf(&buf, "VIDEO_HEIGHT=%d\n", media.VideoHeight)
 	fmt.Fprintf(&buf, "VIDEO_FPS=%d\n", media.VideoFPS)
 	fmt.Fprintf(&buf, "VIDEO_BITRATE=%d\n", media.VideoBitrate)
-	// Disable mic streaming; audio is handled locally for TTS only.
+	audioDevice := audio.CaptureDevice
-	fmt.Fprintf(&buf, "AUDIO_ENABLE=0\n")
+	if audioDevice == "" {
+		audioDevice = "plughw:0,0"
+	}
+	if audio.SampleRate <= 0 {
+		audio.SampleRate = 16000
+	}
+	if audio.Channels <= 0 {
+		audio.Channels = 1
+	}
+	fmt.Fprintf(&buf, "AUDIO_ENABLE=%d\n", boolToInt(audio.CaptureEnabled))
+	fmt.Fprintf(&buf, "AUDIO_DEVICE=%s\n", audioDevice)
+	fmt.Fprintf(&buf, "AUDIO_RATE=%d\n", audio.SampleRate)
+	fmt.Fprintf(&buf, "AUDIO_CHANNELS=%d\n", audio.Channels)
 	if err := os.WriteFile(publisherEnvPath, buf.Bytes(), 0o640); err != nil {
 		return err
 	}

pi/roverd/roverd.sample.yaml

@@ -40,7 +40,7 @@ cameraServo:
 audio:
   captureEnabled: false
   captureDevice: plughw:0,0
-  sampleRate: 48000
+  sampleRate: 16000
   channels: 1
   bitrate: 24000
   ttsEnabled: false