Hi, I am currently working on a project related to audio processing applications on embedded systems. Briefly, I am supposed to develop a hardware accelerator that supports fast Vorbis encoding. For this purpose, I am required to convert all the floating-point calculations to fixed-point as done in the Tremor implementation (however, I need to do the same for encoding). I would like to know if it is possible (since there's no tremor-like implementation for encoding) and if yes, what would be a good way to start doing so. Also, I have read in a couple of papers that Vorbis encoding is not suitable for embedded devices with low-power constraints. Can you please share any information or experience on this subject? Thank you, Mohit.