The 5 Stages of the Roku Audio Decoupling Flow

1. Hardware Demuxing (MPEG-TS): The incoming HLS chunk contains multiplexed Elementary Streams (PES): video (H.264 / HEVC) and audio (AAC-LC, AC3 Dolby Digital, or EAC3 Dolby Digital Plus). The hardware demuxer separates these streams based on their Packet Identifiers (PID).
2. Video Pipeline Acceleration: The video PES stream is fed directly to the dedicated hardware VPU (Video Processing Unit). Decoding 4K 60fps frames takes negligible silicon latency (typically 12ms to 24ms).
3. Audio DSP Downmix Queue: Unlike video, if the audio stream contains 6 discrete surround channels (5.1 EAC3) and the TV’s internal stereo speakers are used, the audio is routed through the TV’s DSP core. The DSP must decode the surround metadata, compute matrix fold-down equations (Left = L + 0.707*C + 0.707*Ls), and apply dynamic range compression (DRC).
4. Roku “Sound Modes” Lookahead Buffer: If features like Automatic Volume Leveling, Night Mode, or Dialogue Enhancement are enabled in Roku OS, the DSP accumulates an audio lookahead FIFO buffer of 150ms to 250ms to anticipate loud volume spikes.
5. Total Cumulative Lag: The video reaches the panel display in ~20ms, while the audio emerges from the DSP in ~320ms, creating a visually jarring 300ms lip-sync lag where mouths move noticeably ahead of speech.

$ ffprobe -show_packets -select_streams a:0 -show_entries packet=pts,dts,pts_time,dts_time live_chunk_482.ts
[PACKET]
codec_type=audio
pts=2849182390
pts_time=31657.582111
dts=2849182390
dts_time=31657.582111
[/PACKET]

$ ffprobe -show_packets -select_streams v:0 -show_entries packet=pts,dts,pts_time,dts_time live_chunk_482.ts
[PACKET]
codec_type=video
pts=2849151240
pts_time=31657.236000
dts=2849147640
dts_time=31657.196000
[/PACKET]

[ANOMALY CALCULATION]:
Video PTS Time: 31,657.236000s
Audio PTS Time: 31,657.582111s
Timestamp Difference (Δ): +0.346111s (+346.1 milliseconds)
[RESULT]: Audio packet timestamps are stamped 346ms BEHIND video frames at encoder origin!

Step 1: Disable Roku DSP “Volume Modes” (Saves 180ms to 250ms)

This single configuration change resolves over 75% of all reported Roku lip-sync complaints:

1. Launch any video stream in your player app (IBO Player Gold, BOB Player, etc.).
2. Press the Star (*) button on your physical Roku remote control to open the Options side-panel.
3. Scroll down to Volume modes (or Sound Settings on Roku TV).
4. Change the setting from Leveling or Night to OFF.
5. Scroll to Speech Clarity and toggle it to OFF.

✓ Disabling Volume Modes flushes the DSP lookahead FIFO buffer, immediately pulling audio forward by up to 250 milliseconds.

Step 2: Force Digital Audio Output to Native PCM Stereo

Force the Roku operating system to bypass surround bitstream negotiation and output clean, zero-latency 2-channel linear PCM:

1. Navigate to the Roku Home screen → Settings → Audio.
2. Select Digital output format (or Audio output).
3. Change from Auto (recommended) to Custom.
4. Set Digital audio format to Stereo.

✓ This forces the Roku SoC to command streaming servers to deliver stereo audio, completely bypassing the heavy Dolby Digital Plus matrix downmixing block.

Step 3: Calibrate HDMI eARC / Soundbar Bitstream Passthrough

If you are outputting sound from your Roku TV to an external Sonos, Bose, or Samsung soundbar over HDMI eARC or Optical Toslink:

• Navigate to Settings → Audio → eARC and ensure eARC is set to ON (regular ARC has limited bandwidth and introduces 80ms latency).
• In Digital output format, select Passthrough. This passes the raw bitstream directly to your soundbar, offloading all decoding computations to the soundbar’s dedicated DSP.
• Under Settings → Audio → TV speakers, toggle internal TV speakers to OFF to prevent acoustic phasing echoes.

sub InitAudioSyncManager(videoNode as Object)
    ' Query Roku device audio capability hardware register
    deviceInfo = CreateObject("roDeviceInfo")
    audioOutputs = deviceInfo.GetAudioOutput()
    
    ' Check if user is connected via HDMI eARC or Optical Toslink
    if deviceInfo.HasFeature("eARC") then
        print "[AUDIO DEBUG]: eARC capability detected, enabling bitstream pass"
        videoNode.audioFormat = "passthrough"
    else
        ' Fallback to low-latency stereo PCM for internal TV speakers
        print "[AUDIO DEBUG]: Forcing low-latency stereo downmix"
        videoNode.audioFormat = "stereo"
    end if

    ' Inject millisecond compensation offset into video player node
    userOffsetMs = RegRead("audio_offset_ms", "playback_prefs", "-250")
    if userOffsetMs <> invalid then
        videoNode.audioDelayMs = userOffsetMs.ToInt()
        print "[AUDIO DEBUG]: Applied audio compensation offset: " + userOffsetMs + "ms"
    end if
end sub

# Lossless Video Passthrough + Low-Latency Audio Downmixing to AAC
ffmpeg -re -i "udp://@239.255.1.1:5000?overrun_nonfatal=1&fifo_size=50000000"   -c:v copy   -c:a aac -b:a 192k -ac 2 -ar 48000   -af "aresample=async=1000:min_hard_comp=0.100000:first_pts=0"   -f hls   -hls_time 2   -hls_list_size 6   -hls_flags delete_segments+append_list+discont_start   -hls_segment_filename "/var/www/hls/live_%03d.ts"   /var/www/hls/live.m3u8