HDMI operation
Audio, video and auxiliary data is transmitted across the three TMDS data channels. The video pixel clock is transmitted on the TMDS clock channel and is used by the receiver as a frequency reference for data recovery on the three TMDS data channels. Video data is carried as a series of 24-bit pixels on the three TMDS data channels. TMDS encoding converts the 8 bits per channel into the 10 bit. The 8-to-10 bit conversion introduces a highly effective error-correction code. Video pixel rates can range from 25 to 165 MHz. Video formats with native pixel rates below 25 Mpixels/s require pixel repetition in order to be carried across a TMDS link. 720 X 480i and 720 X 576i video format timings shall always be pixel repeated.
The video pixels can be encoded in either RGB, YCRCB 4:4:4 or YCRCB 4:2:2 formats. In all three cases, up to 24 bits per pixel can be transferred. While video is transmitted in a serial 10-bit format, audio and auxiliary data are transmitted in a packet structure.
Basic audio functionality consists of a single L-PCM (linear PCM) audio stream at sample rates of 32, 44.1 or 48 kHz. This can accommodate any normal stereo stream. Optionally, HDMI can carry a single such stream at sample rates up to 192 kHz or from two to four such streams (3–8 audio channels) at sample rates up to 96 kHz. HDMI can also carry a compressed (e.g. surround-sound) audio stream at sample rates up to 192 kHz. HDMI can also carry from 2 to 8 channels of one bit audio (1-bit delta-sigma modulated signal stream such as that used by super audio CD).
The HDMI link, which carries both audio and video is driven by a TMDS (video) clock. It does not retain the original audio sample clock. The task of recreating this clock at the receiving end is called audio clock regen- eration. This is made simple since the audio and video clocks are generated in the first place from a common clock (known as coherent clocks).
For a variety of reasons, an HDMI link may add a delay to the audio and/or video. Due to the uneven transmission of audio data, the delay shall be considered to be the average delay of all of the audio sample pack- ets over the course of three steady-state video frames.