Post Go back to editing

AD7480: How to implement of "Lip Sync"

Category: Datasheet/Specs
Product Number: ADV7480

Hi all
My customer development IVI using ADV7480 now.
Could you please answer follow question.

Is it possible to achieve "Auto Lipsync Correction" in the HDMI specifications using the ADV7480?

If possible, please let me know how.

Best Regards
aimPoint

l

Parents
  • Hi,

    Please find the expert comments below.

    When the Audio is extracted from the HDMI, the ADV7840 passes the audio samples through a FIFO. This means that the delay through the Audio will vary as the FIFO fills and empties. Typically, the delay is approximately 70 audio samples (based on a half-full FIFO).

    In system terms, this delay is negligible and will not cause any "Lip sync" problems.

    Thanks,
    Dharani S

  • Hi Dharani S
    Thank you for replay.

    When I checked "HDMI Vendor-Specific Data Block (HDMI VSDB)" in HDMI Specification, "Audio_Latency" was 500ms maximum.

    The Audio FIFO installed in the ADV7480 has 128 data.
    In order to realize "Auto Lipsync Correction", it seems necessary to provide an external delay buffer memory and take system measures.
    Is this understanding correct?

    Best Regards
    AimPoint

Reply
  • Hi Dharani S
    Thank you for replay.

    When I checked "HDMI Vendor-Specific Data Block (HDMI VSDB)" in HDMI Specification, "Audio_Latency" was 500ms maximum.

    The Audio FIFO installed in the ADV7480 has 128 data.
    In order to realize "Auto Lipsync Correction", it seems necessary to provide an external delay buffer memory and take system measures.
    Is this understanding correct?

    Best Regards
    AimPoint

Children
  • Hi,

    As per expert's comments,

    Even though audio maybe running at 48kHz, the audio is packetized and transported over the first available data island in the video stream at video pixel clock rates.  At the sink they are de-packetized and sent to the speakers at 48kHz rate.  

    The various delays stated were based on basic architectural delays. They are not hard numbers. Even if your numbers are correct (66 lines, 1.458ms) which they are not accurate, the human observer cannot detect lipsync issues until the delay exceeds about 10ms. lipsync issues start to occur for scenes with high visual motion and sharp sounds like a gun fire scene were you might see the flash of the gun before you hear the 'bang' from the gun.

    Thanks,
    Dharani S

  • Hi Dharani S


    Thank you for your detailed explanation about Lipsync.
    However, the need for external hardware is not mentioned.

    Since the Audio FIFO built into the ADV7480 has 128 stages,
    in the case of 48kHz sampling, 128/48e03=0.002666,
    and the amount of delay that can be achieved is up to 2.66ms.

    Therefore, in order to implement Lipsync,
    external hardware that delays audio data is required.
    Is it correct?

    Best Regards
    AimPoint

  • Hi,

    As per our expert’s comments, there are interrupts that tell you when the audio PLL locks.

    The part will be outputting audio as soon as the PLL locks, so given the latency - yes you could miss some audio samples. This is pretty normal. You could start capturing video and audio on a VSYNC if you really want to make sure you get complete frames. This is not normally a problem though.

    As mentioned earlier, this delay is negligible and will not cause any "Lip sync" problems.

    Based on your requirement, if you need to introduce a delay in audio data, you can use external hardware devices known as audio delay units or digital signal processors (DSPs).

    Thanks,
    Dharani S

  • Hi Dharani S
    Thank you for your reply. Very helpful.
    I will close this question.

    Best Regards
    AimPoint