Unlike linear PCM systems, neither Dolby Digital nor DTS allocate a fixed numbers of bits to any channel. Instead, Dolby Digital and DTS feed their sub-bands/channels from 'global bit-pools'; the total number of bits allocated to any single channel constantly varies as a result. Sub-bands containing frequencies the human ear is more sensitive to are allocated more bits from the available bit-pool than sub-bands the human ear is less able to detect. Individual frequencies within these sub-bands are allocated data depending on their relative perceptibility when compared to neighbouring frequencies (as determined by the perceptual codes' masking algorithms). In DTS's case, a technique called 'forward-adaptive bit-allocation' is used. Using this technique, the allocation of data to each sub-band is pre-determined exclusively by the encoder. This information is explicitly conveyed to the decoder along with the actual bits to be used. Forward-adaptive bit-allocation's primary advantage is that the psychoacoustic model used resides exclusively within the encoder. Because the model is encoder-based, extremely complex psychoacoustic coding algorithms can be used (as decoder processing ability isn't a limiting factor). Forward-adaptive bit-allocation also allows psychoacoustic model modifications and improvements to be passed directly on to installed decoders, essentially 'future-proofing' DTS decoders from premature obsolescence.
Forward-adaptive bit-allocation's primary drawback is that explicit 'side-information', or 'metadata', is needed to direct and control the decoder's allocation of data to sub-bands; this extra information takes up space that might otherwise have been used for audio reproduction. Dolby Digital uses a hybrid technique incorporating elements of both forward- and backward- adaptive bit-allocation. Like DTS encoders, Dolby Digital encoders must also instruct their decoders to allocate bits to particular sub-bands, but don't need to transmit these instructions with such explicit detail. Dolby Digital decoders already include a very basic 'core' copy of Dolby Digital's perceptual coding algorithm. Because the decoder already 'knows' roughly how the bits should be allocated the encoder only needs to transmit information about specific variations from the decoder's own internal algorithm. Dolby Digital's metadata uses relatively little of the available bandwidth, leaving more data available for audio reproduction (which is a good thing, considering Dolby Digital's bit-pool is considerably smaller than DTS Digital Surround's).
Because DTS's sophistication resides almost entirely within the encoder, DTS decoders are 'passive' and relatively simple. Improvements in the Coherent Acoustics coding system can be passed directly to the decoder, allowing improvements in overall audio performance to be utilised by all decoders, regardless of age. Despite the fact that Dolby Digital decoders contain a fixed core AC-3 algorithm, Dolby Digital is also encoder biased. As such, Dolby Digital decoders can also benefit from advances in the AC-3 coding system. Dolby Digital manages to utilise most of the advantages of both backward-adaptive bit-allocation techniques (decoder-based processing, minimal metadata requirements) and forward-adaptive bit-allocation (encoder-based 'intelligence', fully upgradeable coding), while eliminating many of their respective drawbacks
