Speech and video dual mode gaussian mixture model scoring accelerator

A computer-implemented apparatus for scoring operations, comprising:

• a Gaussian Mixture Model scoring accelerator hardware, the Gaussian Mixture Model scoring accelerator hardware configured to perform scoring operations based at least in part on a buffered plurality of Gaussian Mixture Model means and a buffered plurality of feature vectors;
• a dual use module associated with the Gaussian Mixture Model scoring accelerator hardware, the dual use module configured to:
• intermittently receive image input and speech input from one or more applications, wherein the image input and speech input include a plurality of Gaussian Mixture Model means and a plurality of feature vectors;
• determine when there is a switch between image input and speech input;
• and
• flip-flop a buffer destination for the plurality of Gaussian Mixture Model means and the plurality of feature vectors between a speech specific buffer destination and a image specific buffer destination.

A computer-implemented system for scoring operations, comprising:

• a plurality of Gaussian Mixture Model buffers, the plurality of Gaussian Mixture Model buffers configured to store a plurality of Gaussian Mixture Model means and a plurality of feature vectors;
• a Gaussian Mixture Model scoring accelerator hardware in communication with the plurality of Gaussian Mixture Model buffers, the Gaussian Mixture Model scoring accelerator hardware configured to perform scoring operations based at least in part on a buffered plurality of Gaussian Mixture Model means and a buffered plurality of feature vectors;
• a dual use module associated with the Gaussian Mixture Model scoring accelerator hardware, the dual use module configured to:
• intermittently receive image input and speech input from one or more applications, wherein the image input and speech input include a plurality of Gaussian Mixture Model means and a plurality of feature vectors;
• determine when there is a switch between image input and speech input;
• and
• flip-flop a buffer destination for the plurality of Gaussian Mixture Model means and the plurality of feature vectors between a speech specific buffer destination and a image specific buffer destination.

At least one machine readable medium comprising:

• a plurality of instructions that in response to being executed on a computing device, causes the computing device to perform the following operations, comprising:
• intermittently receiving, via a dual use module, image input and speech input from one or more applications, wherein the image input and speech input include a plurality of Gaussian Mixture Model means and a plurality of feature vectors;
• determining, via the dual use module, when there is a switch between image input and speech input;
• and
• flip-flopping, via the dual use module, a buffer destination for the plurality of Gaussian Mixture Model means and the plurality of feature vectors between a speech specific buffer destination and a image specific buffer destination.