Development plans

Short Term Plans

We expect the following changes in the short term:

• modules for speaker adaptation with CMLLR (Constrained/Feature Space MLLR)

Medium Term Plans

On the longer term (several months), we plan to implement the following:

• a ‘FLaVoR’ (graph based) decoder, see the FLaVoR website for more details

Longer Term Plans

Following items are on our wish list, but have not obtained a clear planning as of this moment:

• a high-level flow-chart scripting language to configure multi-stage recognizers, parallel recognizers (rover alike) with the available processing blocks
• missing data based feature extraction