L64105 LSI Logic Corporation, L64105 Datasheet - Page 344

• Manufacturer Part Number: L64105
• Description: Mpeg-2 Audio/video Decoder
• Manufacturer: LSI Logic Corporation

10.4.2 MPEG Audio Decoding
10-12
MPEG audio encoding is performed by transforming the input signals
from the time domain to the frequency domain and dividing them into 32
frequency subband samples. The subband samples are then quantized,
normalized, and encoded using a variable length encoding scheme. For
decoding, this process is reversed.
Figure 10.3
The host can override the bitstream sample resolution for the decoder by
setting the Overwrite Quantization bit in Register 366
programming the Host Quantization bits in the same register for 20- or
24-bit samples. The decoder extends the samples accordingly.
The host can also scale the output PCM samples down from their
bitstream levels in increments of 1/256 of the levels by writing to Register
362, PCM Scale [7:0],
the audio output.
Audio Decoder Module
Input bitstream parsing: The audio frame is unpacked and parsed,
and the various pieces of coding information are demultiplexed.
Bit allocation decoding: The bit allocation information is decoded first
and is used to parse the scale factors and audio subband samples
later.
Scale factor decoding: The scale-factor index (Layer II only) and
scale factors are unpacked. The 32 audio subband samples are also
parsed using the bit allocation information from the bitstream.
Reconstruction of samples: The subband data samples are
requantized and denormalized.
Subband synthesis: Subband synthesis converts the frequency
domain subband vectors back to time domain PCM samples by
performing the inverse transformation.
Output PCM samples: The decoder transfers 32, 16-bit PCM
samples at a time to be played by the output interface modules.
shows the MPEG audio decoding flow.
page
4-84. A setting of 0x00 in this register mutes
(page
4-89) and