Embedded Speech Transcription and Speaker Identification using the RecoMadeEasy(R) engines. The demo shows the fully embedded speaker diarization on live radio stations with different English dialects. This technology has been created over the past 16 years at Recognition Technologies, Inc. (http://www.recotechnologies.com). The first version of this product was in use in 2003. It may be run on an embedded processor such as a Raspberry PI. It provides realtime speech transcription, speaker segmentation, and speaker identification, together providing a full diarization in most English dialects, using a single model with a 240,000 word vocabulary. This RecoMadeEasy (R) engine may also run on a laptop, a desktop, or a server.
For more information see http://www.RecoTechnologies.com