Speaker recognition using dynamic time warp template spotting

An article comprising a storage medium having stored thereon instructions for identifying a speaker, that when executed by a computing platform, result in:

• computing local distances between test frames and a target template;
• computing accumulated distances from the local distances;
• and
• identifying one or more accumulated distances below a predetermined threshold to identify endpoints of an utterance within the test frames to define a test template that includes a portion of the utterance within the endpoints.

A speaker verification system comprising:

• a memory to store a target template;
• a processing element to compute local distances between test frames and the target template, compute accumulated distances from the local distances, to identify one or more accumulated distances below a predetermined threshold to identify endpoints of an utterance within the test frames to define a test template, and to perform a time warping process on the test template and target template to authenticate a speaker;
• and
• a decision element to grant access to the speaker in response to an authentication by the time warping process.

A pattern recognition system comprising:

• a local distance computing element to compute local distances between test frames and a target template;
• an accumulated distance computing element to compute accumulated distances from the local distances;
• an identifying element to identify one or more accumulated distances below a predetermined threshold identify endpoints of an utterance within the test frames to define a test template;
• and
• a time warping processing element to perform a time warping process on the test template and target template to authenticate the speaker.